euler: use enum and static values for EULER ORDER, remove some macros

this must be more robust and flexible api

.github/FUNDING.yml

@@ -0,0 +1,8 @@
+# These are supported funding model platforms
+
+github: # Replace with up to 4 GitHub Sponsors-enabled usernames e.g., [user1, user2]
+patreon: # Replace with a single Patreon username
+open_collective: cglm
+ko_fi: # Replace with a single Ko-fi username
+tidelift: # Replace with a single Tidelift platform-name/package-name e.g., npm/babel
+custom: # Replace with a single custom sponsorship URL

.gitignore

@@ -51,9 +51,24 @@ cscope.*
 test/*.trs
 test/test_*
 *.log
-test-*
 test/.libs/*
 test/tests
 cglm_arm/*
 cglm_test_ios/*
 cglm_test_iosTests/*
+docs/build/*
+win/cglm_test_*
+* copy.*
+*.o
+*.obj
+*codeanalysis.*.xml
+*codeanalysis.xml
+*.lib
+*.tlog
+win/x64
+win/x85
+win/Debug
+cglm-test-ios*
+/cglm.pc
+test-driver
+Default-568h@2x.png

.gitmodules

@@ -1,3 +0,0 @@
-[submodule "test/lib/cmocka"]
-	path = test/lib/cmocka
-	url = git://git.cryptomilk.org/projects/cmocka.git

.travis.yml

@@ -36,14 +36,10 @@ branches:
   only:
     - master
 
-before_install:
-  - pip install --user cpp-coveralls
-
 script:
-  - sh ./build-deps.sh
   - sh ./autogen.sh
   - if [[ "$CC" == "gcc" && "$CODE_COVERAGE" == "ON" ]]; then
-      ./configure CFLAGS="-ftest-coverage -fprofile-arcs";
+      ./configure CFLAGS="-ftest-coverage -fprofile-arcs -coverage";
     else
       ./configure;
     fi
@@ -52,10 +48,15 @@ script:
 
 after_success:
   - if [[ "$CC" == "gcc" && "$CODE_COVERAGE" == "ON" ]]; then
+      pip install --user cpp-coveralls &&
       coveralls
         --build-root .
         --exclude lib
         --exclude test
         --gcov-options '\-lp'
-        --verbose;
+        --verbose &&
+      bash <(curl -s https://codecov.io/bash);
     fi
+
+# after_failure:
+#   - cat ./test-suite.log

CONTRIBUTING.md

@@ -1,11 +1,11 @@
 # CONTRIBUTING
 
-Any contributions (code, documentation, ...) are welcome. This project uses [cmocka](http://cmocka.org) for testing, you may need to check their documentation 
+Any contributions (code, documentation, ...) are welcome.
 
 # New Features
 - This library may not accept all new features, it is better to create an issue and get approval before coding
 - You must add test for every new feature
-- The feature must be compiled in both UNIX/POSIX systems (e.g. macos, linux...) and Windows
+- The feature must be compiled on both UNIX/POSIX systems (e.g. macos, linux...) and Windows
 
 # Code Style
 This library is written with C99, don't try to add C++ files (yes it can compiled into lib),

CREDITS

@@ -0,0 +1,83 @@
+This library [initially] used some [piece of] implementations
+(may include codes) from these open source projects/resources:
+
+1. Initial Affine Transforms
+The original glm repo (g-truc), url: https://github.com/g-truc/glm
+
+LICENSE[S]:
+  The Happy Bunny License (Modified MIT License)
+  The MIT License
+  Copyright (c) 2005 - 2016 G-Truc Creation
+
+FULL LICENSE: https://github.com/g-truc/glm/blob/master/copying.txt
+
+2. Initial Quaternions
+Anton's OpenGL 4 Tutorials book source code:
+
+LICENSE:
+  OpenGL 4 Example Code.
+  Accompanies written series "Anton's OpenGL 4 Tutorials"
+  Email: anton at antongerdelan dot net
+  First version 27 Jan 2014
+  Copyright Dr Anton Gerdelan, Trinity College Dublin, Ireland.
+
+3. Euler Angles
+  David Eberly
+  Geometric Tools, LLC http://www.geometrictools.com/
+  Copyright (c) 1998-2016. All Rights Reserved.
+
+  Computing Euler angles from a rotation matrix (euler.pdf)
+  Gregory G. Slabaugh
+
+4. Extracting Planes
+Fast Extraction of Viewing Frustum Planes from the World-View-Projection Matrix
+Authors:
+  Gil Gribb (ggribb@ravensoft.com)
+  Klaus Hartmann (k_hartmann@osnabrueck.netsurf.de)
+
+5. Transform AABB
+Transform Axis Aligned Bounding Boxes:
+http://dev.theomader.com/transform-bounding-boxes/
+https://github.com/erich666/GraphicsGems/blob/master/gems/TransBox.c
+
+6. Cull frustum
+http://www.txutxi.com/?p=584
+http://old.cescg.org/CESCG-2002/DSykoraJJelinek/
+
+7. Quaternions
+Initial mat4_quat is borrowed from Apple's simd library
+
+8. Vector Rotation using Quaternion
+https://gamedev.stackexchange.com/questions/28395/rotating-vector3-by-a-quaternion
+
+9. Sphere AABB intersect
+https://github.com/erich666/GraphicsGems/blob/master/gems/BoxSphere.c
+
+10. Horizontal add
+https://stackoverflow.com/questions/6996764/fastest-way-to-do-horizontal-float-vector-sum-on-x86
+
+11. de casteljau implementation and comments
+https://forums.khronos.org/showthread.php/10264-Animations-in-1-4-1-release-notes-revision-A/page2?highlight=bezier
+https://forums.khronos.org/showthread.php/10644-Animation-Bezier-interpolation
+https://forums.khronos.org/showthread.php/10387-2D-Tangents-in-Bezier-Splines?p=34164&viewfull=1#post34164
+https://forums.khronos.org/showthread.php/10651-Animation-TCB-Spline-Interpolation-in-COLLADA?highlight=bezier
+
+12. vec2 cross product
+http://allenchou.net/2013/07/cross-product-of-2d-vectors/
+
+13. Ken Shoemake's algorithm Implementation and Euler
+Ken Shoemake's algorithm impl. is taken from this repo by permission:
+  https://github.com/erich666/GraphicsGems/blob/master/gemsiv/euler_angle
+
+* -------------------------- GraphicsGems EULA ----------------------------- *
+|   http://www.realtimerendering.com/resources/GraphicsGems/                 |
+|                                                                            |
+| EULA: The Graphics Gems code is copyright-protected. In other words, you   |
+| cannot claim the text of the code as your own and resell it. Using the     |
+| code is permitted in any program, product, or library, non-commercial or   |
+| commercial. Giving credit is not required, though is a nice gesture.       |
+| The code comes as-is, and if there are any flaws or problems with any Gems |
+| code, nobody involved with Gems - authors, editors, publishers, or         |
+| webmasters - are to be held responsible. Basically, don't be a jerk, and   |
+| remember that anything free comes with no guarantee.                       |
+* -------------------------------- END --------------------------------------*/

LICENSE

@@ -19,36 +19,3 @@ AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 SOFTWARE.
-
--
-
-This library [initially] used some [piece of] implementations
-(may include codes) from these open source projects/resources:
-
-1. Affine Transforms
-The original glm repo (g-truc), url: https://github.com/g-truc/glm
-
-LICENSE[S]:
-  The Happy Bunny License (Modified MIT License)
-  The MIT License
-  Copyright (c) 2005 - 2016 G-Truc Creation
-
-FULL LICENSE: https://github.com/g-truc/glm/blob/master/copying.txt
-
-2. Quaternions
-Anton's OpenGL 4 Tutorials book source code:
-
-LICENSE:
-  OpenGL 4 Example Code.
-  Accompanies written series "Anton's OpenGL 4 Tutorials"
-  Email: anton at antongerdelan dot net
-  First version 27 Jan 2014
-  Copyright Dr Anton Gerdelan, Trinity College Dublin, Ireland.
-
-3. Euler Angles
-  David Eberly
-  Geometric Tools, LLC http://www.geometrictools.com/
-  Copyright (c) 1998-2016. All Rights Reserved.
-
-  Computing Euler angles from a rotation matrix (euler.pdf)
-  Gregory G. Slabaugh

Makefile.am

@@ -0,0 +1,169 @@
+#******************************************************************************
+# Copyright (c), Recep Aslantas.                                              *
+#                                                                             *
+# MIT License (MIT), http://opensource.org/licenses/MIT                       *
+# Full license can be found in the LICENSE file                               *
+#                                                                             *
+#******************************************************************************
+
+ACLOCAL_AMFLAGS = -I m4
+AM_CFLAGS = -Wall \
+            -std=gnu11 \
+            -O3 \
+            -Wstrict-aliasing=2 \
+            -fstrict-aliasing \
+            -pedantic \
+            -Werror=strict-prototypes
+
+lib_LTLIBRARIES = libcglm.la
+libcglm_la_LDFLAGS = -no-undefined -version-info 0:1:0
+
+checkLDFLAGS = -L./.libs \
+               -lm \
+               -lcglm
+checkCFLAGS = $(AM_CFLAGS) \
+              -I./include
+
+check_PROGRAMS = test/tests
+TESTS = $(check_PROGRAMS)
+
+test_tests_LDFLAGS = $(checkLDFLAGS)
+test_tests_CFLAGS  = $(checkCFLAGS)
+
+cglmdir=$(includedir)/cglm
+cglm_HEADERS = include/cglm/version.h \
+               include/cglm/common.h \
+               include/cglm/types.h \
+               include/cglm/types-struct.h \
+               include/cglm/cglm.h \
+               include/cglm/call.h \
+               include/cglm/struct.h \
+               include/cglm/cam.h \
+               include/cglm/io.h \
+               include/cglm/mat4.h \
+               include/cglm/mat3.h \
+               include/cglm/mat2.h \
+               include/cglm/affine.h \
+               include/cglm/vec2.h \
+               include/cglm/vec2-ext.h \
+               include/cglm/vec3.h \
+               include/cglm/vec3-ext.h \
+               include/cglm/vec4.h \
+               include/cglm/vec4-ext.h \
+               include/cglm/euler.h \
+               include/cglm/util.h \
+               include/cglm/quat.h \
+               include/cglm/affine-mat.h \
+               include/cglm/plane.h \
+               include/cglm/frustum.h \
+               include/cglm/box.h \
+               include/cglm/color.h \
+               include/cglm/project.h \
+               include/cglm/sphere.h \
+               include/cglm/ease.h \
+               include/cglm/curve.h \
+               include/cglm/bezier.h \
+               include/cglm/applesimd.h
+               
+cglm_calldir=$(includedir)/cglm/call
+cglm_call_HEADERS = include/cglm/call/mat4.h \
+                    include/cglm/call/mat3.h \
+                    include/cglm/call/mat2.h \
+                    include/cglm/call/vec2.h \
+                    include/cglm/call/vec3.h \
+                    include/cglm/call/vec4.h \
+                    include/cglm/call/affine.h \
+                    include/cglm/call/io.h \
+                    include/cglm/call/cam.h \
+                    include/cglm/call/quat.h \
+                    include/cglm/call/euler.h \
+                    include/cglm/call/plane.h \
+                    include/cglm/call/frustum.h \
+                    include/cglm/call/box.h \
+                    include/cglm/call/project.h \
+                    include/cglm/call/sphere.h \
+                    include/cglm/call/ease.h \
+                    include/cglm/call/curve.h \
+                    include/cglm/call/bezier.h
+
+cglm_simddir=$(includedir)/cglm/simd
+cglm_simd_HEADERS = include/cglm/simd/intrin.h \
+                    include/cglm/simd/x86.h \
+                    include/cglm/simd/arm.h
+
+cglm_simd_sse2dir=$(includedir)/cglm/simd/sse2
+cglm_simd_sse2_HEADERS = include/cglm/simd/sse2/affine.h \
+                         include/cglm/simd/sse2/mat4.h \
+                         include/cglm/simd/sse2/mat3.h \
+                         include/cglm/simd/sse2/quat.h
+
+cglm_simd_avxdir=$(includedir)/cglm/simd/avx
+cglm_simd_avx_HEADERS = include/cglm/simd/avx/mat4.h \
+                        include/cglm/simd/avx/affine.h
+
+cglm_simd_neondir=$(includedir)/cglm/simd/neon
+cglm_simd_neon_HEADERS = include/cglm/simd/neon/mat4.h
+
+cglm_structdir=$(includedir)/cglm/struct
+cglm_struct_HEADERS = include/cglm/struct/mat4.h \
+                      include/cglm/struct/mat3.h \
+                      include/cglm/struct/vec2.h \
+                      include/cglm/struct/vec2-ext.h \
+                      include/cglm/struct/vec3.h \
+                      include/cglm/struct/vec3-ext.h \
+                      include/cglm/struct/vec4.h \
+                      include/cglm/struct/vec4-ext.h \
+                      include/cglm/struct/affine.h \
+                      include/cglm/struct/io.h \
+                      include/cglm/struct/cam.h \
+                      include/cglm/struct/quat.h \
+                      include/cglm/struct/euler.h \
+                      include/cglm/struct/plane.h \
+                      include/cglm/struct/frustum.h \
+                      include/cglm/struct/box.h \
+                      include/cglm/struct/project.h \
+                      include/cglm/struct/sphere.h \
+                      include/cglm/struct/color.h \
+                      include/cglm/struct/curve.h
+
+libcglm_la_SOURCES=\
+    src/euler.c \
+    src/affine.c \
+    src/io.c \
+    src/quat.c \
+    src/cam.c \
+    src/vec2.c \
+    src/vec3.c \
+    src/vec4.c \
+    src/mat2.c \
+    src/mat3.c \
+    src/mat4.c \
+    src/plane.c \
+    src/frustum.c \
+    src/box.c \
+    src/project.c \
+    src/sphere.c \
+    src/ease.c \
+    src/curve.c \
+    src/bezier.c
+
+test_tests_SOURCES=\
+    test/runner.c \
+    test/src/test_common.c \
+    test/src/tests.c \
+    test/src/test_cam.c \
+    test/src/test_clamp.c \
+    test/src/test_euler.c \
+    test/src/test_bezier.c \
+    test/src/test_struct.c
+
+pkgconfig_DATA=cglm.pc
+
+# When running configure with --prefix, $VPATH references
+# the source directory that post-build.sh is in. When not
+# using a prefix, $VPATH will be unset, so we need to fall
+# back to using . to run the script.
+#export VPATH
+
+# all-local:
+#	sh $${VPATH:-.}/post-build.sh

README.md

@@ -1,26 +1,56 @@
 # 🎥 OpenGL Mathematics (glm) for `C`
 [![Build Status](https://travis-ci.org/recp/cglm.svg?branch=master)](https://travis-ci.org/recp/cglm)
-[![Build status](https://ci.appveyor.com/api/projects/status/av7l3gc0yhfex8y4/branch/master?svg=true)](https://ci.appveyor.com/project/recp/cglm/branch/master)
+ [![Build status](https://ci.appveyor.com/api/projects/status/av7l3gc0yhfex8y4/branch/master?svg=true)](https://ci.appveyor.com/project/recp/cglm/branch/master)
+[![Documentation Status](https://readthedocs.org/projects/cglm/badge/?version=latest)](http://cglm.readthedocs.io/en/latest/?badge=latest)
 [![Coverage Status](https://coveralls.io/repos/github/recp/cglm/badge.svg?branch=master)](https://coveralls.io/github/recp/cglm?branch=master)
+[![codecov](https://codecov.io/gh/recp/cglm/branch/master/graph/badge.svg)](https://codecov.io/gh/recp/cglm)
 [![Codacy Badge](https://api.codacy.com/project/badge/Grade/6a62b37d5f214f178ebef269dc4a6bf1)](https://www.codacy.com/app/recp/cglm?utm_source=github.com&utm_medium=referral&utm_content=recp/cglm&utm_campaign=Badge_Grade)
-
-The original glm library is for C++ only (templates, namespaces, classes...), this library targeted to C99 but currently you can use it for C89 safely by language extensions e.g `__register`
+[![Backers on Open Collective](https://opencollective.com/cglm/backers/badge.svg)](#backers)
+[![Sponsors on Open Collective](https://opencollective.com/cglm/sponsors/badge.svg)](#sponsors)
 
 #### Documentation
 
-Almost all functions (inline versions) and parameters are documented inside related headers. <br />
-Complete documentation is in progress: http://cglm.readthedocs.io
+Almost all functions (inline versions) and parameters are documented inside the corresponding headers. <br />
+Complete documentation: http://cglm.readthedocs.io
 
 #### Note for previous versions:
 
-- _dup (duplicate) is changed to _copy. For instance `glm_vec_dup -> glm_vec_copy`
+- _dup (duplicate) is changed to _copy. For instance `glm_vec_dup -> glm_vec3_copy`
 - OpenGL related functions are dropped to make this lib platform/third-party independent
 - make sure you have latest version and feel free to report bugs, troubles
+- **[bugfix]** euler angles was implemented in reverse order (extrinsic) it was fixed, now they are intrinsic. Make sure that
+you have the latest version
+- **[major change]** by starting v0.4.0, quaternions are stored as [x, y, z, w], it was [w, x, y, z] in v0.3.5 and earlier versions
+- **[api rename]** by starting v0.4.5, **glm_simd** functions are renamed to **glmm_**  
+- **[new option]** by starting v0.4.5, you can disable alignment requirement, check options in docs.  
+- **[major change]** by starting v0.5.0, vec3 functions use **glm_vec3_** namespace, it was **glm_vec_** until v0.5.0
+- **[major change]** by starting v0.5.1, built-in alignment is removed from **vec3** and **mat3** types
 
 #### Note for C++ developers:
-If you don't aware about original GLM library yet, you may also want to look at:
+If you are not aware of the original GLM library yet, you may also want to look at:
 https://github.com/g-truc/glm
 
+#### Note for new comers (Important):
+- `vec4` and `mat4` variables must be aligned. (There will be unaligned versions later)
+- **in** and **[in, out]** parameters must be initialized (please). But **[out]** parameters not, initializing out param is  also redundant
+- All functions are inline if you don't want to use pre-compiled versions with glmc_ prefix, you can ignore build process. Just include headers.
+- if your debugger takes you to cglm headers then make sure you are not trying to copy vec4 to vec3 or alig issues...
+- Welcome!
+
+#### Note for experienced developers:
+- Since I'm testing this library in my projects, sometimes bugs occurs; finding that bug[s] and making improvements would be more easy with multiple developer/contributor and their projects or knowledge. Consider to make some tests if you suspect something is wrong and any feedbacks, contributions and bug reports are always welcome.
+
+#### Allocations?
+`cglm` doesn't alloc any memory on heap. So it doesn't provide any allocator. You should alloc memory for **out** parameters too if you pass pointer of memory location. Don't forget that **vec4** (also quat/**versor**) and **mat4** must be aligned (16-bytes), because *cglm* uses SIMD instructions to optimize most operations if available.
+
+#### Returning vector or matrix... ?
+
+**cglm** supports both *ARRAY API* and *STRUCT API*, so you can return structs if you utilize struct api (`glms_`).
+
+#### Other APIs like Vulkan, Metal, Dx?
+Currently *cglm* uses default clip space configuration (-1, 1) for camera functions (perspective, extract corners...), in the future other clip space configurations will be supported
+
+<hr/>
 
 <table>
   <tbody>
@@ -38,10 +68,11 @@ https://github.com/g-truc/glm
 </table>
 
 ## Features
+- array api and struct api, you can use arrays or structs.
 - general purpose matrix operations (mat4, mat3)
 - chain matrix multiplication (square only)
 - general purpose vector operations (cross, dot, rotate, proj, angle...)
-- affine transforms
+- affine transformations
 - matrix decomposition (extract rotation, scaling factor)
 - optimized affine transform matrices (mul, rigid-body inverse)
 - camera (lookat)
@@ -50,16 +81,25 @@ https://github.com/g-truc/glm
 - euler angles / yaw-pitch-roll to matrix
 - extract euler angles
 - inline or pre-compiled function call
+- frustum (extract view frustum planes, corners...)
+- bounding box  (AABB in Frustum (culling), crop, merge...)
+- bounding sphere
+- project, unproject
+- easing functions
+- curves
+- curve interpolation helpers (S*M*C, deCasteljau...)
+- helpers to convert cglm types to Apple's simd library to pass cglm types to Metal GL without packing them on both sides
+- and others...
 
 <hr />
 
 You have two option to call a function/operation: inline or library call (link)
-Almost all functions are marked inline (always_inline) so compiler probably will inline.
+Almost all functions are marked inline (always_inline) so compiler will probably inline.
 To call pre-compiled version, just use `glmc_` (c stands for 'call') instead of `glm_`.
 
 ```C
-  #include <cglm.h>        /* for inline */
-  #include <cglm-call.h>   /* for library call (this also includes cglm.h) */
+  #include <cglm/cglm.h>   /* for inline */
+  #include <cglm/call.h>   /* for library call (this also includes cglm.h) */
 
   mat4 rot, trans, rt;
   /* ... */
@@ -81,7 +121,7 @@ You can pass matrices and vectors as array to functions rather than get address.
   glm_translate(m, (vec3){1.0f, 0.0f, 0.0f});
 ```
 
-Library contains general purpose mat4 mul and inverse functions but also contains some special form (optimized) of these functions for affine transform matrices. If you want to multiply two affine transform matrices you can use glm_mul instead of glm_mat4_mul and glm_inv_tr (ROT + TR) instead glm_mat4_inv
+Library contains general purpose mat4 mul and inverse functions, and also contains some special forms (optimized) of these functions for affine transformations' matrices. If you want to multiply two affine transformation matrices you can use glm_mul instead of glm_mat4_mul and glm_inv_tr (ROT + TR) instead glm_mat4_inv
 ```C
 /* multiplication */
 mat4 modelMat;
@@ -91,27 +131,48 @@ glm_mul(T, R, modelMat);
 glm_inv_tr(modelMat);
 ```
 
-## License
-MIT. check the LICENSE file
+### Struct API
+
+The struct API works as follows, note the `s` suffix on types, the `glms_` prefix on functions and the `GLMS_` prefix on constants:
+
+```C
+#include <cglm/struct.h>
+
+mat4s mat = GLMS_MAT4_IDENTITY_INIT;
+mat4s inv = glms_mat4_inv(mat);
+```
+
+Struct functions generally take their parameters as *values* and *return* their results, rather than taking pointers and writing to out parameters. That means your parameters can usually be `const`, if you're into that.
+
+The types used are actually unions that allow access to the same data multiple ways. One of those ways involves anonymous structures, available since C11. MSVC also supports it for earlier C versions out of the box and GCC/Clang do if you enable `-fms-extensions`. To explicitly enable these anonymous structures, `#define CGLM_USE_ANONYMOUS_STRUCT` to `1`, to disable them, to `0`. For backward compatibility, you can also `#define CGLM_NO_ANONYMOUS_STRUCT` (value is irrelevant) to disable them. If you don't specify explicitly, cglm will do a best guess based on your compiler and the C version you're using.
 
 ## Build
 
 ### Unix (Autotools)
 
-```text
-$ sh ./build-deps.sh # run only once (dependencies)
-$
+```bash
 $ sh autogen.sh
 $ ./configure
 $ make
-$ make install
-$ [sudo] make install
+$ make check # [Optional]
+$ [sudo] make install # [Optional]
 ```
 
+This will also install pkg-config files so you can use
+`pkg-config --cflags cglm` and `pkg-config --libs cglm` to retrieve compiler
+and linker flags.
+
+The files will be installed into the given prefix (usually `/usr/local` by
+default on Linux), but your pkg-config may not be configured to actually check
+there. You can figure out where it's looking by running `pkg-config --variable
+pc_path pkg-config` and change the path the files are installed to via
+`./configure --with-pkgconfigdir=/your/path`. Alternatively, you can add the
+prefix path to your `PKG_CONFIG_PATH` environment variable.
+
 ### Windows (MSBuild)
-Windows related build files, project files are located in `win` folder,
+Windows related build file and project files are located in `win` folder,
 make sure you are inside `cglm/win` folder.
-Code Analysis are enabled, it may take awhile to build
+Code Analysis is enabled, so it may take awhile to build.
 
 ```Powershell
 $ cd win
@@ -122,45 +183,138 @@ if `msbuild` won't work (because of multi version VS) then try to build with `de
 $ devenv cglm.sln /Build Release
 ```
 
-## How to use
-If you want to use inline versions of funcstions then; include main header
-```C
-#include <cglm.h>
+#### Running Tests on Windows
+
+You can see test project in same visual studio solution file. It is enough to run that project to run tests.
+
+### Building Docs
+First you need install Sphinx: http://www.sphinx-doc.org/en/master/usage/installation.html
+then:
+```bash
+$ cd docs
+$ sphinx-build source build
 ```
-the haeder will include all headers. Then call func you want e.g. rotate vector by axis:
+it will compile docs into build folder, you can run index.html inside that function.
+
+## How to use
+If you want to use the inline versions of functions, then include the main header
 ```C
-glm_vec_rotate(v1, glm_rad(45), (vec3){1.0f, 0.0f, 0.0f});
+#include <cglm/cglm.h>
+```
+the header will include all headers. Then call the func you want e.g. rotate vector by axis:
+```C
+glm_vec3_rotate(v1, glm_rad(45), (vec3){1.0f, 0.0f, 0.0f});
 ```
 some functions are overloaded :) e.g you can normalize vector:
 ```C
-glm_vec_normalize(vec);
+glm_vec3_normalize(vec);
 ```
 this will normalize vec and store normalized vector into `vec` but if you will store normalized vector into another vector do this:
 ```C
-glm_vec_normalize_to(vec, result);
+glm_vec3_normalize_to(vec, result);
 ```
 like this function you may see `_to` postfix, this functions store results to another variables and save temp memory
 
 
 to call pre-compiled versions include header with `c` postfix, c means call. Pre-compiled versions are just wrappers.
 ```C
-#include <cglm-call.h>
+#include <cglm/call.h>
 ```
-this header will include all heaers with c postfix. You need to call functions with c posfix:
+this header will include all headers with c postfix. You need to call functions with c posfix:
 ```C
-glmc_vec_normalize(vec);
+glmc_vec3_normalize(vec);
 ```
 
-Function usage and parameters are documented inside related headers.
+Function usage and parameters are documented inside related headers. You may see same parameter passed twice in some examples like this:
+```C
+glm_mat4_mul(m1, m2, m1);
+
+/* or */
+glm_mat4_mul(m1, m1, m1);
+```
+the first two parameter are **[in]** and the last one is **[out]** parameter. After multiplying *m1* and *m2*, the result is stored in *m1*. This is why we send *m1* twice. You may store the result in a different matrix, this is just an example.
+
+### Example: Computing MVP matrix
+
+#### Option 1
+```C
+mat4 proj, view, model, mvp;
+
+/* init proj, view and model ... */
+
+glm_mat4_mul(proj, view, viewProj);
+glm_mat4_mul(viewProj, model, mvp);
+```
+
+#### Option 2
+```C
+mat4 proj, view, model, mvp;
+
+/* init proj, view and model ... */
+
+glm_mat4_mulN((mat4 *[]){&proj, &view, &model}, 3, mvp);
+```
+
+## How to send matrix to OpenGL
+
+mat4 is array of vec4 and vec4 is array of floats. `glUniformMatrix4fv` functions accecpts `float*` as `value` (last param), so you can cast mat4 to float* or you can pass first column of matrix as beginning of memory of matrix:
+
+Option 1: Send first column
+```C
+glUniformMatrix4fv(location, 1, GL_FALSE, matrix[0]);
+
+/* array of matrices */
+glUniformMatrix4fv(location, 1, GL_FALSE, matrix[0][0]);
+```
+
+Option 2: Cast matrix to pointer type (also valid for multiple dimensional arrays)
+```C
+glUniformMatrix4fv(location, 1, GL_FALSE, (float *)matrix);
+```
+
+You can pass matrices the same way to other APIs e.g. Vulkan, DX...
+
 ## Notes
 
 - This library uses float types only, does not support Integers, Double... yet
 - If headers are not working properly with your compiler, IDE please open an issue, because I'm using GCC and clang to test it maybe sometimes MSVC
 
 **TODO:**
-- [ ] Unit tests
+- [ ] Unit tests (In Progress)
 - [ ] Unit tests for comparing cglm with glm results
 - [x] Add version info
 - [ ] Unaligned operations (e.g. `glm_umat4_mul`)
-- [ ] Extra documentation
-- [ ] ARM Neon Arch
+- [x] Extra documentation
+- [ ] ARM Neon Arch (In Progress)
+
+
+## Contributors
+
+This project exists thanks to all the people who contribute. [[Contribute](CONTRIBUTING.md)].
+<a href="graphs/contributors"><img src="https://opencollective.com/cglm/contributors.svg?width=890&button=false" /></a>
+
+
+## Backers
+
+Thank you to all our backers! 🙏 [[Become a backer](https://opencollective.com/cglm#backer)]
+
+<a href="https://opencollective.com/cglm#backers" target="_blank"><img src="https://opencollective.com/cglm/backers.svg?width=890"></a>
+
+
+## Sponsors
+
+Support this project by becoming a sponsor. Your logo will show up here with a link to your website. [[Become a sponsor](https://opencollective.com/cglm#sponsor)]
+
+<a href="https://opencollective.com/cglm/sponsor/0/website" target="_blank"><img src="https://opencollective.com/cglm/sponsor/0/avatar.svg"></a>
+<a href="https://opencollective.com/cglm/sponsor/1/website" target="_blank"><img src="https://opencollective.com/cglm/sponsor/1/avatar.svg"></a>
+<a href="https://opencollective.com/cglm/sponsor/2/website" target="_blank"><img src="https://opencollective.com/cglm/sponsor/2/avatar.svg"></a>
+<a href="https://opencollective.com/cglm/sponsor/3/website" target="_blank"><img src="https://opencollective.com/cglm/sponsor/3/avatar.svg"></a>
+<a href="https://opencollective.com/cglm/sponsor/4/website" target="_blank"><img src="https://opencollective.com/cglm/sponsor/4/avatar.svg"></a>
+<a href="https://opencollective.com/cglm/sponsor/5/website" target="_blank"><img src="https://opencollective.com/cglm/sponsor/5/avatar.svg"></a>
+<a href="https://opencollective.com/cglm/sponsor/6/website" target="_blank"><img src="https://opencollective.com/cglm/sponsor/6/avatar.svg"></a>
+<a href="https://opencollective.com/cglm/sponsor/7/website" target="_blank"><img src="https://opencollective.com/cglm/sponsor/7/avatar.svg"></a>
+<a href="https://opencollective.com/cglm/sponsor/8/website" target="_blank"><img src="https://opencollective.com/cglm/sponsor/8/avatar.svg"></a>
+<a href="https://opencollective.com/cglm/sponsor/9/website" target="_blank"><img src="https://opencollective.com/cglm/sponsor/9/avatar.svg"></a>
+
+## License
+MIT. check the LICENSE file

autogen.sh

@@ -8,17 +8,14 @@
 
 cd $(dirname "$0")
 
-if [ "$(uname)" = "Darwin" ]; then
-libtoolBin=$(which glibtoolize)
-libtoolBinDir=$(dirname "${libtoolBin}")
-
-if [ ! -f "${libtoolBinDir}/libtoolize" ]; then
-ln -s $libtoolBin "${libtoolBinDir}/libtoolize"
-fi
-fi
-
 autoheader
-libtoolize
+
+if [ "$(uname)" = "Darwin" ]; then
+  glibtoolize
+else
+  libtoolize
+fi
+
 aclocal -I m4
 autoconf
 automake --add-missing --copy

build-deps.sh

@@ -1,30 +0,0 @@
-#! /bin/sh
-#
-# Copyright (c), Recep Aslantas.
-#
-# MIT License (MIT), http://opensource.org/licenses/MIT
-# Full license can be found in the LICENSE file
-#
-
-# check if deps are pulled
-git submodule update --init --recursive
-
-# fix glibtoolize
-
-cd $(dirname "$0")
-
-if [ "$(uname)" = "Darwin" ]; then
-  libtoolBin=$(which glibtoolize)
-  libtoolBinDir=$(dirname "${libtoolBin}")
-  ln -s $libtoolBin "${libtoolBinDir}/libtoolize"
-fi
-
-# general deps: gcc make autoconf automake libtool cmake
-
-# test - cmocka
-cd ./test/lib/cmocka
-mkdir build
-cd build
-cmake -DCMAKE_INSTALL_PREFIX=/usr -DCMAKE_BUILD_TYPE=Debug ..
-make -j8
-cd ../../../../

cglm.pc.in

@@ -0,0 +1,11 @@
+prefix=@prefix@
+exec_prefix=@exec_prefix@
+libdir=@libdir@
+includedir=@includedir@
+
+Name: @PACKAGE_NAME@
+Description: OpenGL Mathematics (glm) for C
+URL: https://github.com/recp/cglm
+Version: @PACKAGE_VERSION@
+Cflags: -I${includedir}
+Libs: -L${libdir} -lcglm @LIBS@

cglm.podspec

@@ -0,0 +1,28 @@
+Pod::Spec.new do |s|
+
+  # Description
+  s.name         = "cglm"
+  s.version      = "0.6.1"
+  s.summary      = "📽 Optimized OpenGL/Graphics Math (glm) for C"
+  s.description  = <<-DESC
+cglm is math library for graphics programming for C. It is similar to original glm but it is written for C instead of C++ (you can use here too). See the documentation or README for all features.
+                   DESC
+
+  s.documentation_url = "http://cglm.readthedocs.io"
+
+  # Home
+  s.homepage     = "https://github.com/recp/cglm"
+  s.license      = { :type => "MIT", :file => "LICENSE" }
+  s.author       = { "Recep Aslantas" => "recp@acm.org" }
+
+  # Sources
+  s.source               = { :git => "https://github.com/recp/cglm.git", :tag => "v#{s.version}" }
+  s.source_files         = "src", "include/cglm/**/*.h"
+  s.public_header_files  = "include", "include/cglm/**/*.h"
+  s.exclude_files        = "src/win/*", "src/dllmain.c", "src/**/*.h"
+  s.preserve_paths       = "include", "src"
+  s.header_mappings_dir  = "include"
+
+  # Linking
+  s.library = "m"
+end

configure.ac

@@ -7,13 +7,30 @@
 #*****************************************************************************
 
 AC_PREREQ([2.69])
-AC_INIT([cglm], [0.2.1], [info@recp.me])
-AM_INIT_AUTOMAKE([-Wall -Werror foreign subdir-objects])
+AC_INIT([cglm], [0.7.1], [info@recp.me])
+AM_INIT_AUTOMAKE([-Wall -Werror foreign subdir-objects serial-tests])
+
+# Don't use the default cflags (-O2 -g), we set ours manually in Makefile.am.
+: ${CFLAGS=""}
 
 AC_CONFIG_MACRO_DIR([m4])
 AC_CONFIG_SRCDIR([src/])
 AC_CONFIG_HEADERS([config.h])
 
+# Dependencies for pkg-config.
+PKG_PROG_PKG_CONFIG
+# Ancient versions of pkg-config (such as the one used in Travis CI)
+# don't have this macro, so we need to do it manually.
+m4_ifdef([PKG_INSTALLDIR], [
+    PKG_INSTALLDIR
+], [
+    AC_ARG_WITH([pkgconfigdir],
+        [AS_HELP_STRING([--with-pkgconfigdir],
+                        [pkg-config installation directory ['${libdir}/pkgconfig']])],,
+        [with_pkgconfigdir=]'${libdir}/pkgconfig')
+    AC_SUBST([pkgconfigdir], [$with_pkgconfigdir])
+])
+
 # Checks for programs.
 AC_PROG_CC
 AM_PROG_CC_C_O
@@ -29,6 +46,7 @@ LT_INIT
 # Checks for libraries.
 AC_CHECK_LIB([m], [floor])
 
+m4_ifdef([AM_SILENT_RULES], [AM_SILENT_RULES([yes])])
 AC_SYS_LARGEFILE
 
 # Checks for header files.
@@ -52,6 +70,6 @@ AC_TYPE_UINT8_T
 # Checks for library functions.
 AC_FUNC_ERROR_AT_LINE
 
-AC_CONFIG_FILES([makefile])
+AC_CONFIG_FILES([Makefile cglm.pc])
 
 AC_OUTPUT

docs/source/affine-mat.rst

@@ -0,0 +1,99 @@
+.. default-domain:: C
+
+affine transform matrix (specialized functions)
+================================================================================
+
+Header: cglm/affine-mat.h
+
+We mostly use glm_mat4_* for 4x4 general and transform matrices. **cglm**
+provides optimized version of some functions. Because affine transform matrix is
+a known format, for instance all last item of first three columns is zero.
+
+You should be careful when using these functions. For instance :c:func:`glm_mul`
+assumes matrix will be this format:
+
+.. code-block:: text
+
+   R  R  R  X
+   R  R  R  Y
+   R  R  R  Z
+   0  0  0  W
+
+if you override zero values here then use :c:func:`glm_mat4_mul` version.
+You cannot use :c:func:`glm_mul` anymore.
+
+Same is also true for :c:func:`glm_inv_tr` if you only have rotation and
+translation then it will work as expected, otherwise you cannot use that.
+
+In the future it may accept scale factors too but currectly it does not.
+
+Table of contents (click func go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Functions:
+
+1. :c:func:`glm_mul`
+#. :c:func:`glm_mul_rot`
+#. :c:func:`glm_inv_tr`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: void  glm_mul(mat4 m1, mat4 m2, mat4 dest)
+
+    | this is similar to glm_mat4_mul but specialized to affine transform
+
+    Matrix format should be:
+
+    .. code-block:: text
+
+       R  R  R  X
+       R  R  R  Y
+       R  R  R  Z
+       0  0  0  W
+
+    this reduces some multiplications. It should be faster than mat4_mul.
+    if you are not sure about matrix format then DON'T use this! use mat4_mul
+
+    Parameters:
+      | *[in]*  **m1**    affine matrix 1
+      | *[in]*  **m2**    affine matrix 2
+      | *[out]* **dest**  result matrix
+
+.. c:function:: void  glm_mul_rot(mat4 m1, mat4 m2, mat4 dest)
+
+    | this is similar to glm_mat4_mul but specialized to rotation matrix
+
+    Right Matrix format should be (left is free):
+
+    .. code-block:: text
+
+       R  R  R  0
+       R  R  R  0
+       R  R  R  0
+       0  0  0  1
+
+    this reduces some multiplications. It should be faster than mat4_mul.
+    if you are not sure about matrix format then DON'T use this! use mat4_mul
+
+    Parameters:
+      | *[in]*  **m1**    affine matrix 1
+      | *[in]*  **m2**    affine matrix 2
+      | *[out]* **dest**  result matrix
+
+.. c:function:: void  glm_inv_tr(mat4 mat)
+
+    | inverse orthonormal rotation + translation matrix (ridig-body)
+
+    .. code-block:: text
+
+       X = | R  T |   X' = | R' -R'T |
+           | 0  1 |        | 0     1 |
+
+    use this if you only have rotation + translation, this should work faster
+    than :c:func:`glm_mat4_inv`
+
+    Don't use this if your matrix includes other things e.g. scale, shear...
+
+    Parameters:
+      | *[in,out]*  **mat**  affine matrix

docs/source/affine.rst

@@ -0,0 +1,343 @@
+.. default-domain:: C
+
+affine transforms
+================================================================================
+
+Header: cglm/affine.h
+
+Initialize Transform Matrices
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+Functions with **_make** prefix expect you don't have a matrix and they create
+a matrix for you. You don't need to pass identity matrix.
+
+But other functions expect you have a matrix and you want to transform them. If
+you didn't have any existing matrix you have to initialize matrix to identity
+before sending to transfrom functions.
+
+There are also functions to decompose transform matrix. These functions can't
+decompose matrix after projected.
+
+Rotation Center
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Rotating functions uses origin as rotation center (pivot/anchor point),
+since scale factors are stored in rotation matrix, same may also true for scalling.
+cglm provides some functions for rotating around at given point e.g.
+**glm_rotate_at**, **glm_quat_rotate_at**. Use them or follow next section for algorihm ("Rotate or Scale around specific Point (Pivot Point / Anchor Point)").
+
+Rotate or Scale around specific Point (Anchor Point)
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+If you want to rotate model around arbibtrary point follow these steps:
+
+1. Move model from pivot point to origin: **translate(-pivot.x, -pivot.y, -pivot.z)**
+2. Apply rotation (or scaling maybe)
+3. Move model back from origin to pivot (reverse of step-1): **translate(pivot.x, pivot.y, pivot.z)**
+
+**glm_rotate_at**, **glm_quat_rotate_at** and their helper functions works that way.
+
+The implementation would be:
+
+.. code-block:: c
+  :linenos:
+
+  glm_translate(m, pivot);
+  glm_rotate(m, angle, axis);
+  glm_translate(m, pivotInv); /* pivotInv = -pivot */
+
+Transforms Order
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+It is important to understand this part especially if you call transform
+functions multiple times
+
+`glm_translate`, `glm_rotate`, `glm_scale` and `glm_quat_rotate` and their
+helpers functions works like this (cglm may provide reverse order too as alternative in the future):
+
+.. code-block:: c
+  :linenos:
+
+  TransformMatrix = TransformMatrix * TraslateMatrix; // glm_translate()
+  TransformMatrix = TransformMatrix * RotateMatrix;   // glm_rotate(), glm_quat_rotate()
+  TransformMatrix = TransformMatrix * ScaleMatrix;    // glm_scale()
+
+As you can see it is multipled as right matrix. For instance what will happen if you call `glm_translate` twice?
+
+.. code-block:: c
+  :linenos:
+
+  glm_translate(transform, translate1); /* transform = transform * translate1 */
+  glm_translate(transform, translate2); /* transform = transform * translate2 */
+  glm_rotate(transform, angle, axis)    /* transform = transform * rotation   */
+
+Now lets try to understand this:
+
+1. You call translate using `translate1` and you expect it will be first transform
+because you call it first, do you?
+
+Result will be **`transform = transform * translate1`**
+
+2. Then you call translate using `translate2` and you expect it will be second transform?
+
+Result will be **`transform = transform * translate2`**. Now lets expand transform,
+it was `transform * translate1` before second call.
+
+Now it is **`transform = transform * translate1 * translate2`**, now do you understand what I say?
+
+3. After last call transform will be:
+
+**`transform = transform * translate1 * translate2 * rotation`**
+
+The order will be; **rotation will be applied first**, then **translate2** then **translate1**
+
+It is all about matrix multiplication order. It is similar to MVP matrix:
+`MVP = Projection * View * Model`, model will be applied first, then view then projection.
+
+**Confused?**
+
+In the end the last function call applied first in shaders.
+
+As alternative way, you can create transform matrices individually then combine manually,
+but don't forget that `glm_translate`, `glm_rotate`, `glm_scale`... are optimized and should be faster (an smaller assembly output) than manual multiplication
+
+.. code-block:: c
+  :linenos:
+
+  mat4 transform1, transform2, transform3, finalTransform;
+
+  glm_translate_make(transform1, translate1);
+  glm_translate_make(transform2, translate2);
+  glm_rotate_make(transform3, angle, axis);
+
+  /* first apply transform1, then transform2, thentransform3 */
+  glm_mat4_mulN((mat4 *[]){&transform3, &transform2, &transform1}, 3, finalTransform);
+
+  /* if you don't want to use mulN, same as above */
+  glm_mat4_mul(transform3, transform2, finalTransform);
+  glm_mat4_mul(finalTransform, transform1, finalTransform);
+
+Now transform1 will be applied first, then transform2 then transform3
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Functions:
+
+1. :c:func:`glm_translate_to`
+#. :c:func:`glm_translate`
+#. :c:func:`glm_translate_x`
+#. :c:func:`glm_translate_y`
+#. :c:func:`glm_translate_z`
+#. :c:func:`glm_translate_make`
+#. :c:func:`glm_scale_to`
+#. :c:func:`glm_scale_make`
+#. :c:func:`glm_scale`
+#. :c:func:`glm_scale_uni`
+#. :c:func:`glm_rotate_x`
+#. :c:func:`glm_rotate_y`
+#. :c:func:`glm_rotate_z`
+#. :c:func:`glm_rotate_make`
+#. :c:func:`glm_rotate`
+#. :c:func:`glm_rotate_at`
+#. :c:func:`glm_rotate_atm`
+#. :c:func:`glm_decompose_scalev`
+#. :c:func:`glm_uniscaled`
+#. :c:func:`glm_decompose_rs`
+#. :c:func:`glm_decompose`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: void  glm_translate_to(mat4 m, vec3 v, mat4 dest)
+
+    translate existing transform matrix by *v* vector and store result in dest
+
+    Parameters:
+      | *[in]*  **m**    affine transfrom
+      | *[in]*  **v**    translate vector [x, y, z]
+      | *[out]* **dest** translated matrix
+
+.. c:function:: void  glm_translate(mat4 m, vec3 v)
+
+    translate existing transform matrix by *v* vector
+    and stores result in same matrix
+
+    Parameters:
+      | *[in, out]* **m**  affine transfrom
+      | *[in]*      **v**  translate vector [x, y, z]
+
+.. c:function:: void  glm_translate_x(mat4 m, float x)
+
+    translate existing transform matrix by x factor
+
+    Parameters:
+      | *[in, out]* **m**  affine transfrom
+      | *[in]*      **v**  x factor
+
+.. c:function:: void  glm_translate_y(mat4 m, float y)
+
+    translate existing transform matrix by *y* factor
+
+    Parameters:
+      | *[in, out]* **m**  affine transfrom
+      | *[in]*      **v**  y factor
+
+.. c:function:: void  glm_translate_z(mat4 m, float z)
+
+    translate existing transform matrix by *z* factor
+
+    Parameters:
+      | *[in, out]* **m**  affine transfrom
+      | *[in]*      **v**  z factor
+
+.. c:function:: void  glm_translate_make(mat4 m, vec3 v)
+
+    creates NEW translate transform matrix by *v* vector.
+
+    Parameters:
+      | *[in, out]* **m**  affine transfrom
+      | *[in]*      **v**  translate vector [x, y, z]
+
+.. c:function:: void  glm_scale_to(mat4 m, vec3 v, mat4 dest)
+
+    scale existing transform matrix by *v* vector and store result in dest
+
+    Parameters:
+      | *[in]*  **m**    affine transfrom
+      | *[in]*  **v**    scale vector [x, y, z]
+      | *[out]* **dest** scaled matrix
+
+.. c:function:: void  glm_scale_make(mat4 m, vec3 v)
+
+    creates NEW scale matrix by v vector
+
+    Parameters:
+      | *[out]* **m** affine transfrom
+      | *[in]*  **v** scale vector [x, y, z]
+
+.. c:function:: void  glm_scale(mat4 m, vec3 v)
+
+    scales existing transform matrix by v vector
+    and stores result in same matrix
+
+    Parameters:
+      | *[in, out]* **m** affine transfrom
+      | *[in]*      **v** scale vector [x, y, z]
+
+.. c:function:: void  glm_scale_uni(mat4 m, float s)
+
+    applies uniform scale to existing transform matrix v = [s, s, s]
+    and stores result in same matrix
+
+    Parameters:
+      | *[in, out]* **m** affine transfrom
+      | *[in]*      **v** scale factor
+
+.. c:function:: void  glm_rotate_x(mat4 m, float angle, mat4 dest)
+
+    rotate existing transform matrix around X axis by angle
+    and store result in dest
+
+    Parameters:
+      | *[in]*  **m**     affine transfrom
+      | *[in]*  **angle** angle (radians)
+      | *[out]* **dest**  rotated matrix
+
+.. c:function:: void  glm_rotate_y(mat4 m, float angle, mat4 dest)
+
+    rotate existing transform matrix around Y axis by angle
+    and store result in dest
+
+    Parameters:
+      | *[in]*  **m**     affine transfrom
+      | *[in]*  **angle** angle (radians)
+      | *[out]* **dest**  rotated matrix
+
+.. c:function:: void  glm_rotate_z(mat4 m, float angle, mat4 dest)
+
+    rotate existing transform matrix around Z axis by angle
+    and store result in dest
+
+    Parameters:
+      | *[in]*  **m**     affine transfrom
+      | *[in]*  **angle** angle (radians)
+      | *[out]* **dest**  rotated matrix
+
+.. c:function:: void  glm_rotate_make(mat4 m, float angle, vec3 axis)
+
+    creates NEW rotation matrix by angle and axis,
+    axis will be normalized so you don't need to normalize it
+
+    Parameters:
+      | *[out]* **m**    affine transfrom
+      | *[in]*  **axis** angle (radians)
+      | *[in]*  **axis** axis
+
+.. c:function:: void  glm_rotate(mat4 m, float angle, vec3 axis)
+
+    rotate existing transform matrix around Z axis by angle and axis
+
+    Parameters:
+      | *[in, out]* **m**     affine transfrom
+      | *[in]*      **angle** angle (radians)
+      | *[in]*      **axis**  axis
+
+.. c:function:: void  glm_rotate_at(mat4 m, vec3 pivot, float angle, vec3 axis)
+
+    rotate existing transform around given axis by angle at given pivot point (rotation center)
+
+    Parameters:
+      | *[in, out]* **m**     affine transfrom
+      | *[in]*      **pivot** pivot, anchor point, rotation center
+      | *[in]*      **angle** angle (radians)
+      | *[in]*      **axis**  axis
+
+.. c:function:: void  glm_rotate_atm(mat4 m, vec3 pivot, float angle, vec3 axis)
+
+    | creates NEW rotation matrix by angle and axis at given point
+    | this creates rotation matrix, it assumes you don't have a matrix
+
+    | this should work faster than glm_rotate_at because it reduces one glm_translate.
+
+    Parameters:
+      | *[in, out]* **m**     affine transfrom
+      | *[in]*      **pivot** pivot, anchor point, rotation center
+      | *[in]*      **angle** angle (radians)
+      | *[in]*      **axis**  axis
+
+.. c:function:: void  glm_decompose_scalev(mat4 m, vec3 s)
+
+    decompose scale vector
+
+    Parameters:
+      | *[in]*  **m**  affine transform
+      | *[out]* **s**  scale vector (Sx, Sy, Sz)
+
+.. c:function:: bool  glm_uniscaled(mat4 m)
+
+    returns true if matrix is uniform scaled.
+    This is helpful for creating normal matrix.
+
+    Parameters:
+      | *[in]*  **m**   matrix
+
+.. c:function:: void  glm_decompose_rs(mat4 m, mat4 r, vec3 s)
+
+    decompose rotation matrix (mat4) and scale vector [Sx, Sy, Sz]
+    DON'T pass projected matrix here
+
+    Parameters:
+      | *[in]*  **m** affine transform
+      | *[out]* **r** rotation matrix
+      | *[out]* **s** scale matrix
+
+.. c:function:: void  glm_decompose(mat4 m, vec4 t, mat4 r, vec3 s)
+
+    decompose affine transform, TODO: extract shear factors.
+    DON'T pass projected matrix here
+
+    Parameters:
+      | *[in]*  **m** affine transfrom
+      | *[out]* **t** translation vector
+      | *[out]* **r** rotation matrix (mat4)
+      | *[out]* **s** scaling vector [X, Y, Z]

docs/source/api.rst

@@ -0,0 +1,53 @@
+API documentation
+================================
+
+Some functions may exist twice,
+once for their namespace and once for global namespace
+to make easier to write very common functions
+
+For instance, in general we use :code:`glm_vec3_dot` to get dot product
+of two **vec3**. Now we can also do this with :code:`glm_dot`,
+same for *_cross* and so on...
+
+The original function stays where it is, the function in global namespace
+of same name is just an alias, so there is no call version of those functions.
+e.g there is no func like :code:`glmc_dot` because *glm_dot* is just alias for
+:code:`glm_vec3_dot`
+
+By including **cglm/cglm.h** header you will include all inline version
+of functions. Since functions in this header[s] are inline you don't need to
+build or link *cglm* against your project.
+
+But by including **cglm/call.h** header you will include all *non-inline*
+version of functions. You need to build *cglm* and link it.
+Follow the :doc:`build` documentation for this
+
+.. toctree::
+   :maxdepth: 1
+   :caption: API categories:
+
+   affine
+   affine-mat
+   cam
+   frustum
+   box
+   quat
+   euler
+   mat4
+   mat3
+   mat2
+   vec3
+   vec3-ext
+   vec4
+   vec4-ext
+   vec2
+   vec2-ext
+   color
+   plane
+   project
+   util
+   io
+   call
+   sphere
+   curve
+   bezier

docs/source/bezier.rst

@@ -0,0 +1,89 @@
+.. default-domain:: C
+
+Bezier
+================================================================================
+
+Header: cglm/bezier.h
+
+Common helpers for cubic bezier and similar curves.
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Functions:
+
+1. :c:func:`glm_bezier`
+2. :c:func:`glm_hermite`
+3. :c:func:`glm_decasteljau`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: float glm_bezier(float s, float p0, float c0, float c1, float p1)
+
+    | cubic bezier interpolation
+    | formula:
+
+    .. code-block:: text
+
+      B(s) = P0*(1-s)^3 + 3*C0*s*(1-s)^2 + 3*C1*s^2*(1-s) + P1*s^3
+
+    | similar result using matrix:
+
+    .. code-block:: text
+
+      B(s) = glm_smc(t, GLM_BEZIER_MAT, (vec4){p0, c0, c1, p1})
+
+    | glm_eq(glm_smc(...), glm_bezier(...)) should return TRUE
+
+    Parameters:
+      | *[in]*  **s**   parameter between 0 and 1
+      | *[in]*  **p0**  begin point
+      | *[in]*  **c0**  control point 1
+      | *[in]*  **c1**  control point 2
+      | *[in]*  **p1**  end point
+
+    Returns:
+        B(s)
+
+.. c:function:: float glm_hermite(float s, float p0, float t0, float t1, float p1)
+
+    | cubic hermite interpolation
+    | formula:
+
+    .. code-block:: text
+
+      H(s) = P0*(2*s^3 - 3*s^2 + 1) + T0*(s^3 - 2*s^2 + s) + P1*(-2*s^3 + 3*s^2) + T1*(s^3 - s^2)
+
+    | similar result using matrix:
+
+    .. code-block:: text
+
+      H(s) = glm_smc(t, GLM_HERMITE_MAT, (vec4){p0, p1, c0, c1})
+
+    | glm_eq(glm_smc(...), glm_hermite(...)) should return TRUE
+
+
+    Parameters:
+      | *[in]*  **s**   parameter between 0 and 1
+      | *[in]*  **p0**  begin point
+      | *[in]*  **t0**  tangent 1
+      | *[in]*  **t1**  tangent 2
+      | *[in]*  **p1**  end point
+
+    Returns:
+        B(s)
+
+.. c:function:: float glm_decasteljau(float prm, float p0, float c0, float c1, float p1)
+
+    | iterative way to solve cubic equation
+
+    Parameters:
+      | *[in]*  **prm** parameter between 0 and 1
+      | *[in]*  **p0**  begin point
+      | *[in]*  **c0**  control point 1
+      | *[in]*  **c1**  control point 2
+      | *[in]*  **p1**  end point
+
+    Returns:
+        parameter to use in cubic equation

docs/source/box.rst

@@ -0,0 +1,181 @@
+.. default-domain:: C
+
+axis aligned bounding box (AABB)
+================================================================================
+
+Header: cglm/box.h
+
+Some convenient functions provided for AABB.
+
+**Definition of box:**
+
+cglm defines box as two dimensional array of vec3.
+The first element is **min** point and the second one is **max** point.
+If you have another type e.g. struct or even another representation then you must
+convert it before and after call cglm box function.
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Functions:
+
+1. :c:func:`glm_aabb_transform`
+#. :c:func:`glm_aabb_merge`
+#. :c:func:`glm_aabb_crop`
+#. :c:func:`glm_aabb_crop_until`
+#. :c:func:`glm_aabb_frustum`
+#. :c:func:`glm_aabb_invalidate`
+#. :c:func:`glm_aabb_isvalid`
+#. :c:func:`glm_aabb_size`
+#. :c:func:`glm_aabb_radius`
+#. :c:func:`glm_aabb_center`
+#. :c:func:`glm_aabb_aabb`
+#. :c:func:`glm_aabb_sphere`
+#. :c:func:`glm_aabb_point`
+#. :c:func:`glm_aabb_contains`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: void  glm_aabb_transform(vec3 box[2], mat4 m, vec3 dest[2])
+
+    | apply transform to Axis-Aligned Bounding Box
+
+    Parameters:
+      | *[in]*  **box**   bounding box
+      | *[in]*  **m**     transform matrix
+      | *[out]* **dest**  transformed bounding box
+
+.. c:function:: void  glm_aabb_merge(vec3 box1[2], vec3 box2[2], vec3 dest[2])
+
+    | merges two AABB bounding box and creates new one
+
+    two box must be in same space, if one of box is in different space then
+    you should consider to convert it's space by glm_box_space
+
+    Parameters:
+      | *[in]*  **box1** bounding box 1
+      | *[in]*  **box2** bounding box 2
+      | *[out]* **dest** merged bounding box
+
+.. c:function:: void  glm_aabb_crop(vec3 box[2], vec3 cropBox[2], vec3 dest[2])
+
+    | crops a bounding box with another one.
+
+    this could be useful for gettng a bbox which fits with view frustum and
+    object bounding boxes. In this case you crop view frustum box with objects
+    box
+
+    Parameters:
+      | *[in]*  **box**      bounding box 1
+      | *[in]*  **cropBox**  crop box
+      | *[out]* **dest**     cropped bounding box
+
+.. c:function:: void  glm_aabb_crop_until(vec3 box[2], vec3 cropBox[2], vec3 clampBox[2], vec3 dest[2])
+
+    | crops a bounding box with another one.
+
+    this could be useful for gettng a bbox which fits with view frustum and
+    object bounding boxes. In this case you crop view frustum box with objects
+    box
+
+    Parameters:
+      | *[in]*  **box**      bounding box
+      | *[in]*  **cropBox**  crop box
+      | *[in]*  **clampBox** miniumum box
+      | *[out]* **dest**     cropped bounding box
+
+.. c:function:: bool  glm_aabb_frustum(vec3 box[2], vec4 planes[6])
+
+    | check if AABB intersects with frustum planes
+
+    this could be useful for frustum culling using AABB.
+
+    OPTIMIZATION HINT:
+       if planes order is similar to LEFT, RIGHT, BOTTOM, TOP, NEAR, FAR
+       then this method should run even faster because it would only use two
+       planes if object is not inside the two planes
+       fortunately cglm extracts planes as this order! just pass what you got!
+
+    Parameters:
+      | *[in]*   **box**     bounding box
+      | *[out]*  **planes**  frustum planes
+
+.. c:function:: void  glm_aabb_invalidate(vec3 box[2])
+
+    | invalidate AABB min and max values
+
+    | It fills *max* values with -FLT_MAX and *min* values with +FLT_MAX
+
+    Parameters:
+      | *[in, out]*   **box**     bounding box
+
+.. c:function:: bool  glm_aabb_isvalid(vec3 box[2])
+
+    | check if AABB is valid or not
+
+    Parameters:
+      | *[in]*   **box**     bounding box
+
+    Returns:
+      returns true if aabb is valid otherwise false
+
+.. c:function:: float  glm_aabb_size(vec3 box[2])
+
+    | distance between of min and max
+
+    Parameters:
+      | *[in]*   **box**     bounding box
+
+    Returns:
+      distance between min - max
+
+.. c:function:: float  glm_aabb_radius(vec3 box[2])
+
+    | radius of sphere which surrounds AABB
+
+    Parameters:
+      | *[in]*   **box**     bounding box
+
+.. c:function:: void  glm_aabb_center(vec3 box[2], vec3 dest)
+
+    | computes center point of AABB
+
+    Parameters:
+      | *[in]*    **box**      bounding box
+      | *[out]*   **dest**     center of bounding box
+
+.. c:function:: bool  glm_aabb_aabb(vec3 box[2], vec3 other[2])
+
+    | check if two AABB intersects
+
+    Parameters:
+      | *[in]*    **box**     bounding box
+      | *[out]*   **other**   other bounding box
+
+.. c:function:: bool  glm_aabb_sphere(vec3 box[2], vec4 s)
+
+    | check if AABB intersects with sphere
+
+    | https://github.com/erich666/GraphicsGems/blob/master/gems/BoxSphere.c
+    | Solid Box - Solid Sphere test.
+
+    Parameters:
+      | *[in]*    **box**     solid bounding box
+      | *[out]*   **s**       solid sphere
+
+.. c:function:: bool  glm_aabb_point(vec3 box[2], vec3 point)
+
+    | check if point is inside of AABB
+
+    Parameters:
+      | *[in]*    **box**     bounding box
+      | *[out]*   **point**   point
+
+.. c:function:: bool  glm_aabb_contains(vec3 box[2], vec3 other[2])
+
+    | check if AABB contains other AABB
+
+    Parameters:
+      | *[in]*    **box**     bounding box
+      | *[out]*   **other**   other bounding box

docs/source/build.rst

@@ -1,24 +1,18 @@
-Building cglm
+Build cglm
 ================================
 
-| **cglm** does not have external dependencies except for unit testing.
-| When you pulled cglm repo with submodules all dependencies will be pulled too.
-| `build-deps.sh` will pull all dependencies/submodules and build for you.
-
-External dependencies:
-  * cmocka - for unit testing
+| **cglm** does not have any external dependencies.
 
 **NOTE:**
 If you only need to inline versions, you don't need to build **cglm**, you don't need to link it to your program.
 Just import cglm to your project as dependency / external lib by copy-paste then use it as usual
 
-**Unix (Autotools):**
+Unix (Autotools):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 .. code-block:: bash
   :linenos:
 
-  $ sh ./build-deps.sh    # run this only once (dependencies)
-
   $ sh autogen.sh
   $ ./configure
   $ make
@@ -26,11 +20,12 @@ Just import cglm to your project as dependency / external lib by copy-paste then
   $ [sudo] make install   # install to system (optional)
 
 **make** will build cglm to **.libs** sub folder in project folder.
-If you don't want to install cglm to your system's folder you can get static and dynamic libs in this folder.
+If you don't want to install **cglm** to your system's folder you can get static and dynamic libs in this folder.
 
-**Build dependencies (windows):**
+Windows (MSBuild):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
-Windows related build files, project files are located in win folder,
+Windows related build files, project files are located in `win` folder,
 make sure you are inside in cglm/win folder.
 
 Code Analysis are enabled, it may take awhile to build.
@@ -50,3 +45,18 @@ then try to build with *devenv*:
   $ devenv cglm.sln /Build Release
 
 Currently tests are not available on Windows.
+
+Documentation (Sphinx):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+**cglm** uses sphinx framework for documentation, it allows lot of formats for documentation. To see all options see sphinx build page:
+
+https://www.sphinx-doc.org/en/master/man/sphinx-build.html
+
+Example build:
+
+.. code-block:: bash
+  :linenos:
+
+  $ cd cglm/docs
+  $ sphinx-build source build

docs/source/call.rst

@@ -0,0 +1,19 @@
+.. default-domain:: C
+
+precompiled functions (call)
+================================================================================
+
+All funcitons in **glm_** namespace are forced to **inline**.
+Most functions also have pre-compiled version.
+
+Precompiled versions are in **glmc_** namespace. *c* in the namespace stands for
+"call".
+
+Since precompiled functions are just wrapper for inline verisons,
+these functions are not documented individually.
+It would be duplicate documentation also it
+would be hard to sync documentation between inline and call verison for me.
+
+By including **clgm/cglm.h** you include all inline verisons. To get precompiled
+versions you need to include **cglm/call.h** header it also includes all
+call versions plus *clgm/cglm.h* (inline verisons)

docs/source/cam.rst

@@ -0,0 +1,313 @@
+.. default-domain:: C
+
+camera
+======
+
+Header: cglm/cam.h
+
+There are many convenient functions for camera. For instance :c:func:`glm_look`
+is just wrapper for :c:func:`glm_lookat`. Sometimes you only have direction
+instead of target, so that makes easy to build view matrix using direction.
+There is also :c:func:`glm_look_anyup` function which can help build view matrix
+without providing UP axis. It uses :c:func:`glm_vec3_ortho` to get a UP axis and
+builds view matrix.
+
+You can also *_default* versions of ortho and perspective to build projection
+fast if you don't care specific projection values.
+
+*_decomp* means decompose; these function can help to decompose projection
+matrices.
+
+ **NOTE**: Be careful when working with high range (very small near, very large
+ far) projection matrices. You may not get exact value you gave.
+ **float** type cannot store very high precision so you will lose precision.
+ Also your projection matrix will be inaccurate due to losing precision
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Functions:
+
+1. :c:func:`glm_frustum`
+#. :c:func:`glm_ortho`
+#. :c:func:`glm_ortho_aabb`
+#. :c:func:`glm_ortho_aabb_p`
+#. :c:func:`glm_ortho_aabb_pz`
+#. :c:func:`glm_ortho_default`
+#. :c:func:`glm_ortho_default_s`
+#. :c:func:`glm_perspective`
+#. :c:func:`glm_persp_move_far`
+#. :c:func:`glm_perspective_default`
+#. :c:func:`glm_perspective_resize`
+#. :c:func:`glm_lookat`
+#. :c:func:`glm_look`
+#. :c:func:`glm_look_anyup`
+#. :c:func:`glm_persp_decomp`
+#. :c:func:`glm_persp_decompv`
+#. :c:func:`glm_persp_decomp_x`
+#. :c:func:`glm_persp_decomp_y`
+#. :c:func:`glm_persp_decomp_z`
+#. :c:func:`glm_persp_decomp_far`
+#. :c:func:`glm_persp_decomp_near`
+#. :c:func:`glm_persp_fovy`
+#. :c:func:`glm_persp_aspect`
+#. :c:func:`glm_persp_sizes`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: void  glm_frustum(float left, float right, float bottom, float top, float nearVal, float farVal, mat4  dest)
+
+    | set up perspective peprojection matrix
+
+    Parameters:
+      | *[in]*  **left**      viewport.left
+      | *[in]*  **right**     viewport.right
+      | *[in]*  **bottom**    viewport.bottom
+      | *[in]*  **top**       viewport.top
+      | *[in]*  **nearVal**   near clipping plane
+      | *[in]*  **farVal**    far clipping plane
+      | *[out]* **dest**      result matrix
+
+.. c:function:: void  glm_ortho(float left, float right, float bottom, float top, float nearVal, float farVal, mat4  dest)
+
+    | set up orthographic projection matrix
+
+    Parameters:
+      | *[in]*  **left**      viewport.left
+      | *[in]*  **right**     viewport.right
+      | *[in]*  **bottom**    viewport.bottom
+      | *[in]*  **top**       viewport.top
+      | *[in]*  **nearVal**   near clipping plane
+      | *[in]*  **farVal**    far clipping plane
+      | *[out]* **dest**      result matrix
+
+.. c:function:: void  glm_ortho_aabb(vec3 box[2], mat4 dest)
+
+    | set up orthographic projection matrix using bounding box
+    | bounding box (AABB) must be in view space
+
+    Parameters:
+      | *[in]*  **box**   AABB
+      | *[in]*  **dest**  result matrix
+
+.. c:function:: void  glm_ortho_aabb_p(vec3 box[2], float padding, mat4 dest)
+
+    | set up orthographic projection matrix using bounding box
+    | bounding box (AABB) must be in view space
+
+    this version adds padding to box
+
+    Parameters:
+      | *[in]*  **box**      AABB
+      | *[in]*  **padding**  padding
+      | *[out]* **d**        result matrix
+
+.. c:function:: void  glm_ortho_aabb_pz(vec3 box[2], float padding, mat4 dest)
+
+    | set up orthographic projection matrix using bounding box
+    | bounding box (AABB) must be in view space
+
+    this version adds Z padding to box
+
+    Parameters:
+      | *[in]*  **box**      AABB
+      | *[in]*  **padding**  padding for near and far
+      | *[out]* **d**        result matrix
+
+    Returns:
+      square of norm / magnitude
+
+.. c:function:: void  glm_ortho_default(float aspect, mat4  dest)
+
+    | set up unit orthographic projection matrix
+
+    Parameters:
+      | *[in]*  **aspect** aspect ration ( width / height )
+      | *[out]* **dest**   result matrix
+
+.. c:function:: void  glm_ortho_default_s(float aspect, float size, mat4  dest)
+
+    | set up orthographic projection matrix with given CUBE size
+
+    Parameters:
+      | *[in]*  **aspect** aspect ration ( width / height )
+      | *[in]*  **size**   cube size
+      | *[out]* **dest**   result matrix
+
+.. c:function:: void  glm_perspective(float fovy, float aspect, float nearVal, float farVal, mat4  dest)
+
+    | set up perspective projection matrix
+
+    Parameters:
+      | *[in]*  **fovy**    field of view angle
+      | *[in]*  **aspect**  aspect ratio ( width / height )
+      | *[in]*  **nearVal** near clipping plane
+      | *[in]*  **farVal**  far clipping planes
+      | *[out]* **dest**    result matrix
+
+.. c:function:: void  glm_persp_move_far(mat4 proj, float deltaFar)
+
+    | extend perspective projection matrix's far distance
+
+    | this function does not guarantee far >= near, be aware of that!
+
+    Parameters:
+      | *[in, out]*  **proj**      projection matrix to extend
+      | *[in]*       **deltaFar**  distance from existing far (negative to shink)
+
+.. c:function:: void glm_perspective_default(float aspect, mat4 dest)
+
+     | set up perspective projection matrix with default near/far
+       and angle values
+
+    Parameters:
+      | *[in]*  **aspect** aspect aspect ratio ( width / height )
+      | *[out]* **dest**   result matrix
+
+.. c:function:: void  glm_perspective_resize(float aspect, mat4 proj)
+
+    | resize perspective matrix by aspect ratio ( width / height )
+      this makes very easy to resize proj matrix when window / viewport reized
+
+    Parameters:
+      | *[in]*      **aspect** aspect ratio ( width / height )
+      | *[in, out]* **proj**   perspective projection matrix
+
+.. c:function:: void  glm_lookat(vec3 eye, vec3 center, vec3 up, mat4 dest)
+
+    | set up view matrix
+
+    **NOTE:** The UP vector must not be parallel to the line of sight from the eye point to the reference point.
+
+    Parameters:
+      | *[in]*  **eye**     eye vector
+      | *[in]*  **center**  center vector
+      | *[in]*  **up**      up vector
+      | *[out]* **dest**    result matrix
+
+.. c:function:: void  glm_look(vec3 eye, vec3 dir, vec3 up, mat4 dest)
+
+    | set up view matrix
+
+    convenient wrapper for :c:func:`glm_lookat`: if you only have direction not
+    target self then this might be useful. Because you need to get target
+    from direction.
+
+    **NOTE:** The UP vector must not be parallel to the line of sight from the eye point to the reference point.
+
+    Parameters:
+      | *[in]*  **eye**     eye vector
+      | *[in]*  **center**  direction vector
+      | *[in]*  **up**      up vector
+      | *[out]* **dest**    result matrix
+
+.. c:function:: void  glm_look_anyup(vec3 eye, vec3 dir, mat4 dest)
+
+    | set up view matrix
+
+    convenient wrapper for :c:func:`glm_look` if you only have direction
+    and if you don't care what UP vector is then this might be useful
+    to create view matrix
+
+    Parameters:
+      | *[in]*  **eye**     eye vector
+      | *[in]*  **center**  direction vector
+      | *[out]* **dest**    result matrix
+
+.. c:function:: void  glm_persp_decomp(mat4 proj, float *nearVal, float *farVal, float *top, float *bottom, float *left, float *right)
+
+    | decomposes frustum values of perspective projection.
+
+    Parameters:
+      | *[in]*  **eye**       perspective projection matrix
+      | *[out]*  **nearVal**  near
+      | *[out]*  **farVal**   far
+      | *[out]*  **top**      top
+      | *[out]*  **bottom**   bottom
+      | *[out]*  **left**     left
+      | *[out]* **right**     right
+
+.. c:function:: void  glm_persp_decompv(mat4 proj, float dest[6])
+
+    | decomposes frustum values of perspective projection.
+    | this makes easy to get all values at once
+
+    Parameters:
+      | *[in]*   **proj**  perspective projection matrix
+      | *[out]*  **dest**  array
+
+.. c:function:: void  glm_persp_decomp_x(mat4 proj, float *left, float *right)
+
+    | decomposes left and right values of perspective projection.
+    | x stands for x axis (left / right axis)
+
+    Parameters:
+      | *[in]*   **proj**   perspective projection matrix
+      | *[out]*  **left**   left
+      | *[out]*  **right**  right
+
+.. c:function:: void  glm_persp_decomp_y(mat4 proj, float *top, float *bottom)
+
+    | decomposes top and bottom values of perspective projection.
+    | y stands for y axis (top / botom axis)
+
+    Parameters:
+      | *[in]*   **proj**    perspective projection matrix
+      | *[out]*  **top**     top
+      | *[out]*  **bottom**  bottom
+
+.. c:function:: void  glm_persp_decomp_z(mat4 proj, float *nearVal, float *farVal)
+
+    | decomposes near and far values of perspective projection.
+    | z stands for z axis (near / far axis)
+
+    Parameters:
+      | *[in]*   **proj**     perspective projection matrix
+      | *[out]*  **nearVal**  near
+      | *[out]*  **farVal**   far
+
+.. c:function:: void  glm_persp_decomp_far(mat4 proj, float * __restrict farVal)
+
+    | decomposes far value of perspective projection.
+
+    Parameters:
+      | *[in]*  **proj**    perspective projection matrix
+      | *[out]* **farVal**  far
+
+.. c:function:: void  glm_persp_decomp_near(mat4 proj, float * __restrict nearVal)
+
+    | decomposes near value of perspective projection.
+
+    Parameters:
+      | *[in]*  **proj**    perspective projection matrix
+      | *[out]* **nearVal** near
+
+.. c:function:: float  glm_persp_fovy(mat4 proj)
+
+    | returns field of view angle along the Y-axis (in radians)
+
+    if you need to degrees, use glm_deg to convert it or use this:
+    fovy_deg = glm_deg(glm_persp_fovy(projMatrix))
+
+    Parameters:
+      | *[in]*  **proj**  perspective projection matrix
+
+    Returns:
+      | fovy in radians
+
+.. c:function:: float  glm_persp_aspect(mat4 proj)
+
+    | returns aspect ratio of perspective projection
+
+    Parameters:
+      | *[in]*  **proj**  perspective projection matrix
+
+.. c:function:: void  glm_persp_sizes(mat4 proj, float fovy, vec4 dest)
+
+    | returns sizes of near and far planes of perspective projection
+
+    Parameters:
+      | *[in]*  **proj**  perspective projection matrix
+      | *[in]*  **fovy**  fovy (see brief)
+      | *[out]* **dest**  sizes order: [Wnear, Hnear, Wfar, Hfar]

docs/source/color.rst

@@ -0,0 +1,34 @@
+.. default-domain:: C
+
+color
+================================================================================
+
+Header: cglm/color.h
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Functions:
+
+1. :c:func:`glm_luminance`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: float  glm_luminance(vec3 rgb)
+
+    | averages the color channels into one value
+
+    This function uses formula in COLLADA 1.5 spec which is
+
+    .. code-block:: text
+
+       luminance = (color.r * 0.212671) +
+                   (color.g * 0.715160) +
+                   (color.b * 0.072169)
+
+    It is based on the ISO/CIE color standards (see ITU-R Recommendation BT.709-4),
+    that averages the color channels into one value
+
+    Parameters:
+      | *[in]*  **rgb**   RGB color

docs/source/conf.py

@@ -30,7 +30,15 @@
 # Add any Sphinx extension module names here, as strings. They can be
 # extensions coming with Sphinx (named 'sphinx.ext.*') or your custom
 # ones.
-extensions = []
+extensions = [
+    'sphinx.ext.doctest',
+    'sphinx.ext.todo',
+    'sphinx.ext.coverage',
+    'sphinx.ext.mathjax',
+    'sphinx.ext.ifconfig',
+    'sphinx.ext.viewcode',
+    'sphinx.ext.githubpages'
+]
 
 # Add any paths that contain templates here, relative to this directory.
 templates_path = ['_templates']
@@ -54,9 +62,9 @@ author = u'Recep Aslantas'
 # built documents.
 #
 # The short X.Y version.
-version = u'0.2.1'
+version = u'0.7.1'
 # The full version, including alpha/beta/rc tags.
-release = u'0.2.1'
+release = u'0.7.1'
 
 # The language for content autogenerated by Sphinx. Refer to documentation
 # for a list of supported languages.
@@ -82,7 +90,7 @@ todo_include_todos = False
 # The theme to use for HTML and HTML Help pages.  See the documentation for
 # a list of builtin themes.
 #
-html_theme = 'alabaster'
+html_theme = 'sphinx_rtd_theme'
 
 # Theme options are theme-specific and customize the look and feel of a theme
 # further.  For a list of options available for each theme, see the
@@ -91,13 +99,13 @@ html_theme = 'alabaster'
 # html_theme_options = {}
 
 html_theme_options = {
-    'github_banner': 'true',
-    'github_button': 'true',
-    'github_user': 'recp',
-    'github_repo': 'cglm',
-    'travis_button': 'true',
-    'show_related': 'true',
-    'fixed_sidebar': 'true'
+    # 'github_banner': 'true',
+    # 'github_button': 'true',
+    # 'github_user': 'recp',
+    # 'github_repo': 'cglm',
+    # 'travis_button': 'true',
+    # 'show_related': 'true',
+    # 'fixed_sidebar': 'true'
 }
 
 # Add any paths that contain custom static files (such as style sheets) here,
@@ -161,3 +169,31 @@ texinfo_documents = [
      author, 'cglm', 'One line description of project.',
      'Miscellaneous'),
 ]
+
+# -- Options for Epub output -------------------------------------------------
+
+# Bibliographic Dublin Core info.
+epub_title = project
+epub_author = author
+epub_publisher = author
+epub_copyright = copyright
+
+# The unique identifier of the text. This can be a ISBN number
+# or the project homepage.
+#
+# epub_identifier = ''
+
+# A unique identification for the text.
+#
+# epub_uid = ''
+
+# A list of files that should not be packed into the epub file.
+epub_exclude_files = ['search.html']
+
+
+# -- Extension configuration -------------------------------------------------
+
+# -- Options for todo extension ----------------------------------------------
+
+# If true, `todo` and `todoList` produce output, else they produce nothing.
+todo_include_todos = True

docs/source/curve.rst

@@ -0,0 +1,41 @@
+.. default-domain:: C
+
+Curve
+================================================================================
+
+Header: cglm/curve.h
+
+Common helpers for common curves. For specific curve see its header/doc
+e.g bezier
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Functions:
+
+1. :c:func:`glm_smc`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: float  glm_smc(float s, mat4 m, vec4 c)
+
+    | helper function to calculate **S** * **M** * **C** multiplication for curves
+
+    | this function does not encourage you to use SMC, instead it is a helper if you use SMC.
+
+    | if you want to specify S as vector then use more generic glm_mat4_rmc() func.
+
+    | Example usage:
+
+    .. code-block:: c
+
+       Bs = glm_smc(s, GLM_BEZIER_MAT, (vec4){p0, c0, c1, p1})
+
+    Parameters:
+      | *[in]*  **s**  parameter between 0 and 1 (this will be [s3, s2, s, 1])
+      | *[in]*  **m**  basis matrix
+      | *[out]* **c**  position/control vector
+
+    Returns:
+        scalar value e.g. Bs

docs/source/euler.rst

@@ -0,0 +1,182 @@
+.. default-domain:: C
+
+euler angles
+============
+
+Header: cglm/euler.h
+
+You may wonder what **glm_euler_sq** type ( **_sq** stands for sequence ) and
+:c:func:`glm_euler_by_order` do.
+I used them to convert euler angles in one coordinate system to another. For
+instance if you have **Z_UP** euler angles and if you want to convert it
+to **Y_UP** axis then :c:func:`glm_euler_by_order` is your friend. For more
+information check :c:func:`glm_euler_order` documentation
+
+You must pass arrays as array, if you use C compiler then you can use something
+like this:
+
+.. code-block:: c
+
+   float pitch, yaw, roll;
+   mat4  rot;
+
+   /* pitch = ...; yaw = ...; roll = ... */
+   glm_euler((vec3){pitch, yaw, roll}, rot);
+
+Rotation Conveniention
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+Current *cglm*'s euler functions uses these convention:
+
+* Tait–Bryan angles (x-y-z convention)
+* Intrinsic rotations (pitch, yaw and roll).
+  This is reserve order of extrinsic (elevation, heading and bank) rotation
+* Right hand rule (actually all rotations in *cglm* use **RH**)
+* All angles used in *cglm* are **RADIANS** not degrees
+
+
+**NOTE**: The default :c:func:`glm_euler` function is the short name of
+:c:func:`glm_euler_xyz` this is why you can't see :c:func:`glm_euler_xyz`.
+When you see an euler function which doesn't have any X, Y, Z suffix then
+assume that uses **_xyz** (or instead it accept order as parameter).
+
+If rotation doesn't work properly, your options:
+
+1. If you use (or paste) degrees convert it to radians before calling an euler function
+
+.. code-block:: c
+
+   float pitch, yaw, roll;
+   mat4  rot;
+
+   /* pitch = degrees; yaw = degrees; roll = degrees */
+   glm_euler((vec3){glm_rad(pitch), glm_rad(yaw), glm_rad(roll)}, rot);
+
+2. Convention mismatch. You may have extrinsic angles,
+   if you do (if you must) then consider to use reverse order e.g if you have
+   **xyz** extrinsic then use **zyx**
+
+3. *cglm* may implemented it wrong, consider to create an issue to report it
+   or pull request to fix it
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Types:
+
+1. glm_euler_sq
+
+Functions:
+
+1. :c:func:`glm_euler_order`
+#. :c:func:`glm_euler_angles`
+#. :c:func:`glm_euler`
+#. :c:func:`glm_euler_xyz`
+#. :c:func:`glm_euler_zyx`
+#. :c:func:`glm_euler_zxy`
+#. :c:func:`glm_euler_xzy`
+#. :c:func:`glm_euler_yzx`
+#. :c:func:`glm_euler_yxz`
+#. :c:func:`glm_euler_by_order`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: glm_euler_sq  glm_euler_order(int ord[3])
+
+    | packs euler angles order to glm_euler_sq enum.
+
+    To use :c:func:`glm_euler_by_order` function you need *glm_euler_sq*. You
+    can get it with this function.
+
+    You can build param like this:
+
+    | X = 0, Y = 1, Z = 2
+
+    if you have ZYX order then you pass this: [2, 1, 0] = ZYX.
+    if you have YXZ order then you pass this: [1, 0, 2] = YXZ
+
+    As you can see first item specifies which axis will be first then the
+    second one specifies which one will be next an so on.
+
+    Parameters:
+      | *[in]*  **ord**  euler angles order [Angle1, Angle2, Angle2]
+
+    Returns:
+      packed euler order
+
+.. c:function:: void  glm_euler_angles(mat4 m, vec3 dest)
+
+    | extract euler angles (in radians) using xyz order
+
+    Parameters:
+      | *[in]*  **m**     affine transform
+      | *[out]* **dest**  angles vector [x, y, z]
+
+.. c:function:: void  glm_euler(vec3 angles, mat4 dest)
+
+    | build rotation matrix from euler angles
+
+    this is alias of glm_euler_xyz function
+
+    Parameters:
+      | *[in]*  **angles**  angles as vector [Xangle, Yangle, Zangle]
+      | *[in]*  **dest**    rotation matrix
+
+.. c:function:: void  glm_euler_xyz(vec3 angles, mat4 dest)
+
+    | build rotation matrix from euler angles
+
+    Parameters:
+      | *[in]*  **angles**  angles as vector [Xangle, Yangle, Zangle]
+      | *[in]*  **dest**    rotation matrix
+
+.. c:function:: void  glm_euler_zyx(vec3 angles, mat4 dest)
+
+    | build rotation matrix from euler angles
+
+    Parameters:
+      | *[in]*  **angles**  angles as vector [Xangle, Yangle, Zangle]
+      | *[in]*  **dest**    rotation matrix
+
+.. c:function:: void  glm_euler_zxy(vec3 angles, mat4 dest)
+
+    | build rotation matrix from euler angles
+
+    Parameters:
+      | *[in]*  **angles**  angles as vector [Xangle, Yangle, Zangle]
+      | *[in]*  **dest**    rotation matrix
+
+.. c:function:: void  glm_euler_xzy(vec3 angles, mat4 dest)
+
+    | build rotation matrix from euler angles
+
+    Parameters:
+      | *[in]*  **angles**  angles as vector [Xangle, Yangle, Zangle]
+      | *[in]*  **dest**    rotation matrix
+
+.. c:function:: void  glm_euler_yzx(vec3 angles, mat4 dest)
+
+    build rotation matrix from euler angles
+
+    Parameters:
+      | *[in]*  **angles**  angles as vector [Xangle, Yangle, Zangle]
+      | *[in]*  **dest**    rotation matrix
+
+.. c:function:: void  glm_euler_yxz(vec3 angles, mat4 dest)
+
+    | build rotation matrix from euler angles
+
+    Parameters:
+      | *[in]*  **angles**  angles as vector [Xangle, Yangle, Zangle]
+      | *[in]*  **dest**    rotation matrix
+
+.. c:function:: void glm_euler_by_order(vec3 angles, glm_euler_sq ord, mat4 dest)
+
+    | build rotation matrix from euler angles with given euler order.
+
+    Use :c:func:`glm_euler_order` function to build *ord* parameter
+
+    Parameters:
+      | *[in]*  **angles**  angles as vector [Xangle, Yangle, Zangle]
+      | *[in]*  **ord**     euler order
+      | *[in]*  **dest**    rotation matrix

docs/source/features.rst

@@ -0,0 +1,23 @@
+Features
+================================================================================
+
+* general purpose matrix operations (mat4, mat3)
+* chain matrix multiplication (square only)
+* general purpose vector operations (cross, dot, rotate, proj, angle...)
+* affine transforms
+* matrix decomposition (extract rotation, scaling factor)
+* optimized affine transform matrices (mul, rigid-body inverse)
+* camera (lookat)
+* projections (ortho, perspective)
+* quaternions
+* euler angles / yaw-pitch-roll to matrix
+* extract euler angles
+* inline or pre-compiled function call
+* frustum (extract view frustum planes, corners...)
+* bounding box (AABB in Frustum (culling), crop, merge...)
+* bounding sphere
+* project, unproject
+* easing functions
+* curves
+* curve interpolation helpers (SMC, deCasteljau...)
+* and other...

docs/source/frustum.rst

@@ -0,0 +1,168 @@
+.. default-domain:: C
+
+frustum
+=============
+
+Header: cglm/frustum.h
+
+cglm provides convenient functions to extract frustum planes, corners...
+All extracted corners are **vec4** so you must create array of **vec4**
+not **vec3**. If you want to store them to save space you msut convert them
+yourself.
+
+**vec4** is used to speed up functions need to corners. This is why frustum
+fucntions use *vec4* instead of *vec3*
+
+Currenty related-functions use [-1, 1] clip space configuration to extract
+corners but you can override it by prodiving **GLM_CUSTOM_CLIPSPACE** macro.
+If you provide it then you have to all bottom macros as *vec4*
+
+Current configuration:
+
+.. code-block:: c
+
+   /* near */
+   GLM_CSCOORD_LBN {-1.0f, -1.0f, -1.0f, 1.0f}
+   GLM_CSCOORD_LTN {-1.0f,  1.0f, -1.0f, 1.0f}
+   GLM_CSCOORD_RTN { 1.0f,  1.0f, -1.0f, 1.0f}
+   GLM_CSCOORD_RBN { 1.0f, -1.0f, -1.0f, 1.0f}
+
+   /* far */
+   GLM_CSCOORD_LBF {-1.0f, -1.0f,  1.0f, 1.0f}
+   GLM_CSCOORD_LTF {-1.0f,  1.0f,  1.0f, 1.0f}
+   GLM_CSCOORD_RTF { 1.0f,  1.0f,  1.0f, 1.0f}
+   GLM_CSCOORD_RBF { 1.0f, -1.0f,  1.0f, 1.0f}
+
+
+Explain of short names:
+  * **LBN**: left  bottom near
+  * **LTN**: left  top    near
+  * **RTN**: right top    near
+  * **RBN**: right bottom near
+  * **LBF**: left  bottom far
+  * **LTF**: left  top    far
+  * **RTF**: right top    far
+  * **RBF**: right bottom far
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Macros:
+
+.. code-block:: c
+
+   GLM_LBN    0 /* left  bottom near */
+   GLM_LTN    1 /* left  top    near */
+   GLM_RTN    2 /* right top    near */
+   GLM_RBN    3 /* right bottom near */
+
+   GLM_LBF    4 /* left  bottom far  */
+   GLM_LTF    5 /* left  top    far  */
+   GLM_RTF    6 /* right top    far  */
+   GLM_RBF    7 /* right bottom far  */
+
+   GLM_LEFT   0
+   GLM_RIGHT  1
+   GLM_BOTTOM 2
+   GLM_TOP    3
+   GLM_NEAR   4
+   GLM_FAR    5
+
+Functions:
+
+1. :c:func:`glm_frustum_planes`
+#. :c:func:`glm_frustum_corners`
+#. :c:func:`glm_frustum_center`
+#. :c:func:`glm_frustum_box`
+#. :c:func:`glm_frustum_corners_at`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: void  glm_frustum_planes(mat4 m, vec4 dest[6])
+
+    | extracts view frustum planes
+
+    planes' space:
+     - if m = proj:     View Space
+     - if m = viewProj: World Space
+     - if m = MVP:      Object Space
+
+    You probably want to extract planes in world space so use viewProj as m
+    Computing viewProj:
+
+    .. code-block:: c
+
+       glm_mat4_mul(proj, view, viewProj);
+
+    Exracted planes order: [left, right, bottom, top, near, far]
+
+    Parameters:
+      | *[in]*  **m**     matrix
+      | *[out]* **dest**  exracted view frustum planes
+
+.. c:function:: void  glm_frustum_corners(mat4 invMat, vec4 dest[8])
+
+    | extracts view frustum corners using clip-space coordinates
+
+    corners' space:
+     - if m = invViewProj: World Space
+     - if m = invMVP:      Object Space
+
+    You probably want to extract corners in world space so use **invViewProj**
+    Computing invViewProj:
+
+    .. code-block:: c
+
+       glm_mat4_mul(proj, view, viewProj);
+       ...
+       glm_mat4_inv(viewProj, invViewProj);
+
+    if you have a near coord at **i** index,
+    you can get it's far coord by i + 4;
+    follow example below to understand that
+
+    For instance to find center coordinates between a near and its far coord:
+
+    .. code-block:: c
+
+       for (j = 0; j < 4; j++) {
+         glm_vec3_center(corners[i], corners[i + 4], centerCorners[i]);
+       }
+
+    corners[i + 4] is far of corners[i] point.
+
+    Parameters:
+      | *[in]*  **invMat** matrix
+      | *[out]* **dest**   exracted view frustum corners
+
+.. c:function:: void  glm_frustum_center(vec4 corners[8], vec4 dest)
+
+    | finds center of view frustum
+
+    Parameters:
+      | *[in]*  **corners**  view frustum corners
+      | *[out]* **dest**     view frustum center
+
+.. c:function:: void  glm_frustum_box(vec4 corners[8], mat4 m, vec3 box[2])
+
+    | finds bounding box of frustum relative to given matrix e.g. view mat
+
+    Parameters:
+      | *[in]*  **corners**  view frustum corners
+      | *[in]*  **m**        matrix to convert existing conners
+      | *[out]* **box**      bounding box as array [min, max]
+
+.. c:function:: void  glm_frustum_corners_at(vec4  corners[8], float splitDist, float farDist, vec4  planeCorners[4])
+
+    | finds planes corners which is between near and far planes (parallel)
+
+    this will be helpful if you want to split a frustum e.g. CSM/PSSM. This will
+    find planes' corners but you will need to one more plane.
+    Actually you have it, it is near, far or created previously with this func ;)
+
+    Parameters:
+      | *[in]*   **corners**       frustum corners
+      | *[in]*   **splitDist**     split distance
+      | *[in]*   **farDist**       far distance (zFar)
+      | *[out]*  **planeCorners**  plane corners [LB, LT, RT, RB]

docs/source/getting_started.rst

@@ -1,40 +1,79 @@
 Getting Started
 ================================
 
+Types:
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
 **cglm** uses **glm** prefix for all functions e.g. glm_lookat. You can see supported types in common header file:
 
 .. code-block:: c
   :linenos:
 
-   typedef float vec3[3];
-   typedef int  ivec3[3];
-   typedef CGLM_ALIGN(16) float vec4[4];
+  typedef float                   vec2[2];
+  typedef float                   vec3[3];
+  typedef int                    ivec3[3];
+  typedef CGLM_ALIGN_IF(16) float vec4[4];
+  typedef vec4                    versor;
+  typedef vec3                    mat3[3];
 
-   typedef vec3 mat3[3];
-   typedef vec4 mat4[4];
-
-   typedef vec4 versor;
+  #ifdef __AVX__
+  typedef CGLM_ALIGN_IF(32) vec4  mat4[4];
+  #else
+  typedef CGLM_ALIGN_IF(16) vec4  mat4[4];
+  #endif
 
 As you can see types don't store extra informations in favor of space.
 You can send these values e.g. matrix to OpenGL directly without casting or calling a function like *value_ptr*
 
-*vec4* and *mat4* requires 16 byte aligment because vec4 and mat4 operations are
-vectorized by SIMD instructions (SSE/AVX).
+Alignment Is Required:
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
-**NOTE:** Unaligned vec4 and unaligned mat4 operations will be supported in the future. Check todo list.
+**vec4** and **mat4** requires 16 (32 for **mat4** if AVX is enabled) byte alignment because **vec4** and **mat4** operations are vectorized by SIMD instructions (SSE/AVX/NEON).
+
+**UPDATE:**
+  By starting v0.4.5 cglm provides an option to disable alignment requirement, it is enabled as default
+
+  | Check :doc:`opt` page for more details
+
+  Also alignment is disabled for older msvc verisons as default. Now alignment is only required in Visual Studio 2017 version 15.6+ if CGLM_ALL_UNALIGNED macro is not defined.
+
+Allocations:
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+*cglm* doesn't alloc any memory on heap. So it doesn't provide any allocator.
+You must allocate memory yourself. You should alloc memory for out parameters too if you pass pointer of memory location. When allocating memory, don't forget that **vec4** and **mat4** require alignment.
+
+**NOTE:** Unaligned **vec4** and unaligned **mat4** operations will be supported in the future. Check todo list.
 Because you may want to multiply a CGLM matrix with external matrix.
 There is no guarantee that non-CGLM matrix is aligned. Unaligned types will have *u* prefix e.g. **umat4**
 
+Array vs Struct:
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+*cglm* uses arrays for vector and matrix types. So you can't access individual
+elements like vec.x, vec.y, vec.z... You must use subscript to access vector elements
+e.g. vec[0], vec[1], vec[2].
+
+Also I think it is more meaningful to access matrix elements with subscript
+e.g **matrix[2][3]** instead of **matrix._23**. Since matrix is array of vectors,
+vectors are also defined as array. This makes types homogeneous.
+
+**Return arrays?**
+
+Since C doesn't support return arrays, cglm also doesn't support this feature.
+
+Function design:
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. image:: cglm-intro.png
+   :width: 492px 
+   :height: 297px
+   :align: center
+
 cglm provides a few way to call a function to do same operation.
 
 * Inline - *glm_, glm_u*
-   * aligned
-   * unaligned (todo)
 * Pre-compiled - *glmc_, glmc_u*
-   * aligned
-   * unaligned (todo)
 
-For instance **glm_mat4_mul** is inline (all *glm_* functions are inline), to make it non-inline (pre-compiled)
+For instance **glm_mat4_mul** is inline (all *glm_* functions are inline), to make it non-inline (pre-compiled),
 call it as **glmc_mat4_mul** from library, to use unaligned version use **glm_umat4_mul** (todo).
 
 Most functions have **dest** parameter for output. For instance mat4_mul func looks like this:

docs/source/index.rst

@@ -3,39 +3,48 @@
    You can adapt this file completely to your liking, but it should at least
    contain the root `toctree` directive.
 
-Welcome to cglm's documentation!
+cglm Documentation
 ================================
 
 **cglm** is optimized 3D math library written in C99 (compatible with C89).
 It is similar to original **glm** library except this is mainly for **C**
 
-This library stores matrices as row-major order but in the future column-major
+This library stores matrices as column-major order but in the future row-major
 is considered to be supported as optional.
 
 Also currently only **float** type is supported for most operations.
 
-**Features**
+.. toctree::
+   :maxdepth: 2
+   :caption: Getting Started:
 
-* general purpose matrix operations (mat4, mat3)
-* chain matrix multiplication (square only)
-* general purpose vector operations (cross, dot, rotate, proj, angle...)
-* affine transforms
-* matrix decomposition (extract rotation, scaling factor)
-* optimized affine transform matrices (mul, rigid-body inverse)
-* camera (lookat)
-* projections (ortho, perspective)
-* quaternions
-* euler angles / yaw-pitch-roll to matrix
-* extract euler angles
-* inline or pre-compiled function call
-* more features (todo)
+   features
+   build
+   getting_started
 
 .. toctree::
    :maxdepth: 2
-   :caption: Table Of Contents:
+   :caption: How To:
 
-   build
-   getting_started
+   opengl
+
+.. toctree::
+   :maxdepth: 2
+   :caption: API:
+
+   api
+
+.. toctree::
+   :maxdepth: 2
+   :caption: Options:
+
+   opt
+
+.. toctree::
+   :maxdepth: 2
+   :caption: Troubleshooting:
+
+   troubleshooting
 
 Indices and tables
 ==================

docs/source/io.rst

@@ -0,0 +1,102 @@
+.. default-domain:: C
+
+io (input / output e.g. print)
+================================================================================
+
+Header: cglm/io.h
+
+There are some built-in print functions which may save your time,
+especially for debugging.
+
+All functions accept **FILE** parameter which makes very flexible.
+You can even print it to file on disk.
+
+In general you will want to print them to console to see results.
+You can use **stdout** and **stderr** to write results to console.
+Some programs may occupy **stdout** but you can still use **stderr**.
+Using **stderr** is suggested.
+
+Example to print mat4 matrix:
+
+.. code-block:: c
+
+   mat4 transform;
+   /* ... */
+   glm_mat4_print(transform, stderr);
+
+**NOTE:** print functions use **%0.4f** precision if you need more
+(you probably will in some cases), you can change it temporary.
+cglm may provide precision parameter in the future
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Functions:
+
+1. :c:func:`glm_mat4_print`
+#. :c:func:`glm_mat3_print`
+#. :c:func:`glm_vec4_print`
+#. :c:func:`glm_vec3_print`
+#. :c:func:`glm_ivec3_print`
+#. :c:func:`glm_versor_print`
+#. :c:func:`glm_aabb_print`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: void  glm_mat4_print(mat4 matrix, FILE * __restrict ostream)
+
+    | print mat4 to given stream
+
+    Parameters:
+      | *[in]*  **matrix**   matrix
+      | *[in]*  **ostream**  FILE to write
+
+.. c:function:: void  glm_mat3_print(mat3 matrix, FILE * __restrict ostream)
+
+    | print mat3 to given stream
+
+    Parameters:
+      | *[in]*  **matrix**   matrix
+      | *[in]*  **ostream**  FILE to write
+
+.. c:function:: void  glm_vec4_print(vec4 vec, FILE * __restrict ostream)
+
+    | print vec4 to given stream
+
+    Parameters:
+      | *[in]*  **vec**      vector
+      | *[in]*  **ostream**  FILE to write
+
+.. c:function:: void  glm_vec3_print(vec3 vec, FILE * __restrict ostream)
+
+    | print vec3 to given stream
+
+    Parameters:
+      | *[in]*  **vec**      vector
+      | *[in]*  **ostream**  FILE to write
+
+.. c:function:: void  glm_ivec3_print(ivec3 vec, FILE * __restrict ostream)
+
+    | print ivec3 to given stream
+
+    Parameters:
+      | *[in]*  **vec**      vector
+      | *[in]*  **ostream**  FILE to write
+
+.. c:function:: void  glm_versor_print(versor vec, FILE * __restrict ostream)
+
+    | print quaternion to given stream
+
+    Parameters:
+      | *[in]*  **vec**      quaternion
+      | *[in]*  **ostream**  FILE to write
+
+.. c:function:: void  glm_aabb_print(versor vec, const char * __restrict tag, FILE * __restrict ostream)
+
+    | print aabb to given stream
+
+    Parameters:
+      | *[in]*  **vec**      aabb (axis-aligned bounding box)
+      | *[in]*  **tag**      tag to find it more easly in logs
+      | *[in]*  **ostream**  FILE to write

docs/source/mat2.rst

@@ -0,0 +1,179 @@
+.. default-domain:: C
+
+mat2
+====
+
+Header: cglm/mat2.h
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Macros:
+
+1. GLM_mat2_IDENTITY_INIT
+#. GLM_mat2_ZERO_INIT
+#. GLM_mat2_IDENTITY
+#. GLM_mat2_ZERO
+
+Functions:
+
+1. :c:func:`glm_mat2_copy`
+#. :c:func:`glm_mat2_identity`
+#. :c:func:`glm_mat2_identity_array`
+#. :c:func:`glm_mat2_zero`
+#. :c:func:`glm_mat2_mul`
+#. :c:func:`glm_mat2_transpose_to`
+#. :c:func:`glm_mat2_transpose`
+#. :c:func:`glm_mat2_mulv`
+#. :c:func:`glm_mat2_scale`
+#. :c:func:`glm_mat2_det`
+#. :c:func:`glm_mat2_inv`
+#. :c:func:`glm_mat2_trace`
+#. :c:func:`glm_mat2_swap_col`
+#. :c:func:`glm_mat2_swap_row`
+#. :c:func:`glm_mat2_rmc`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: void glm_mat2_copy(mat2 mat, mat2 dest)
+
+    copy mat2 to another one (dest).
+
+    Parameters:
+      | *[in]*  **mat**   source
+      | *[out]* **dest**  destination
+
+.. c:function:: void glm_mat2_identity(mat2 mat)
+
+    copy identity mat2 to mat, or makes mat to identiy
+
+    Parameters:
+      | *[out]* **mat**  matrix
+
+.. c:function:: void glm_mat2_identity_array(mat2 * __restrict mat, size_t count)
+
+    make given matrix array's each element identity matrix
+
+    Parameters:
+      | *[in,out]* **mat**  matrix array (must be aligned (16/32) if alignment is not disabled)
+      | *[in]* **count**  count of matrices
+
+.. c:function:: void glm_mat2_zero(mat2 mat)
+
+    make given matrix zero
+
+    Parameters:
+      | *[in,out]* **mat**  matrix to
+
+.. c:function:: void glm_mat2_mul(mat2 m1, mat2 m2, mat2 dest)
+
+    multiply m1 and m2 to dest
+    m1, m2 and dest matrices can be same matrix, it is possible to write this:
+
+    .. code-block:: c
+
+       mat2 m = GLM_mat2_IDENTITY_INIT;
+       glm_mat2_mul(m, m, m);
+
+    Parameters:
+      | *[in]*  **m1**    left matrix
+      | *[in]*  **m2**    right matrix
+      | *[out]* **dest**  destination matrix
+
+.. c:function:: void glm_mat2_transpose_to(mat2 m, mat2 dest)
+
+    transpose mat4 and store in dest
+    source matrix will not be transposed unless dest is m
+
+    Parameters:
+      | *[in]*  **mat**   source
+      | *[out]* **dest**  destination
+
+.. c:function:: void glm_mat2_transpose(mat2 m)
+
+    tranpose mat2 and store result in same matrix
+
+    Parameters:
+      | *[in]*  **mat**   source
+      | *[out]* **dest**  destination
+
+.. c:function:: void glm_mat2_mulv(mat2 m, vec2 v, vec2 dest)
+
+    multiply mat4 with vec4 (column vector) and store in dest vector
+
+    Parameters:
+      | *[in]*  **mat**   mat2 (left)
+      | *[in]*  **v**     vec2 (right, column vector)
+      | *[out]* **dest**  destination (result, column vector)
+
+.. c:function:: void  glm_mat2_scale(mat2 m, float s)
+
+    multiply matrix with scalar
+
+    Parameters:
+      | *[in, out]* **mat**   matrix
+      | *[in]*      **dest**  scalar
+
+.. c:function:: float  glm_mat2_det(mat2 mat)
+
+    returns mat2 determinant
+
+    Parameters:
+      | *[in]*  **mat**   matrix
+
+    Returns:
+        mat2 determinant
+
+.. c:function:: void glm_mat2_inv(mat2 mat, mat2 dest)
+
+    inverse mat2 and store in dest
+
+    Parameters:
+      | *[in]*  **mat**  matrix
+      | *[out]* **dest** destination (inverse matrix)
+
+.. c:function:: void glm_mat2_trace(mat2 m)
+
+    | sum of the elements on the main diagonal from upper left to the lower right
+
+    Parameters:
+      | *[in]*  **m**  matrix
+
+    Returns:
+        trace of matrix
+
+.. c:function:: void glm_mat2_swap_col(mat2 mat, int col1, int col2)
+
+    swap two matrix columns
+
+    Parameters:
+      | *[in, out]*  **mat**   matrix
+      | *[in]*       **col1**  col1
+      | *[in]*       **col2**  col2
+
+.. c:function:: void glm_mat2_swap_row(mat2 mat, int row1, int row2)
+
+    swap two matrix rows
+
+    Parameters:
+      | *[in, out]*  **mat**   matrix
+      | *[in]*       **row1**  row1
+      | *[in]*       **row2**  row2
+
+.. c:function:: float glm_mat2_rmc(vec2 r, mat2 m, vec2 c)
+
+    | **rmc** stands for **Row** * **Matrix** * **Column**
+
+    | helper for  R (row vector) * M (matrix) * C (column vector)
+
+    | the result is scalar because R * M = Matrix1x2 (row vector),
+    | then Matrix1x2 * Vec2 (column vector) = Matrix1x1 (Scalar)
+
+    Parameters:
+      | *[in]*  **r**  row vector or matrix1x2
+      | *[in]*  **m**  matrix2x2
+      | *[in]*  **c**  column vector or matrix2x1
+
+    Returns:
+        scalar value e.g. Matrix1x1

docs/source/mat3.rst

@@ -0,0 +1,189 @@
+.. default-domain:: C
+
+mat3
+====
+
+Header: cglm/mat3.h
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Macros:
+
+1. GLM_MAT3_IDENTITY_INIT
+#. GLM_MAT3_ZERO_INIT
+#. GLM_MAT3_IDENTITY
+#. GLM_MAT3_ZERO
+#. glm_mat3_dup(mat, dest)
+
+Functions:
+
+1. :c:func:`glm_mat3_copy`
+#. :c:func:`glm_mat3_identity`
+#. :c:func:`glm_mat3_identity_array`
+#. :c:func:`glm_mat3_zero`
+#. :c:func:`glm_mat3_mul`
+#. :c:func:`glm_mat3_transpose_to`
+#. :c:func:`glm_mat3_transpose`
+#. :c:func:`glm_mat3_mulv`
+#. :c:func:`glm_mat3_quat`
+#. :c:func:`glm_mat3_scale`
+#. :c:func:`glm_mat3_det`
+#. :c:func:`glm_mat3_inv`
+#. :c:func:`glm_mat3_trace`
+#. :c:func:`glm_mat3_swap_col`
+#. :c:func:`glm_mat3_swap_row`
+#. :c:func:`glm_mat3_rmc`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: void  glm_mat3_copy(mat3 mat, mat3 dest)
+
+    copy mat3 to another one (dest).
+
+    Parameters:
+      | *[in]*  **mat**   source
+      | *[out]* **dest**  destination
+
+.. c:function:: void  glm_mat3_identity(mat3 mat)
+
+    copy identity mat3 to mat, or makes mat to identiy
+
+    Parameters:
+      | *[out]* **mat**  matrix
+
+.. c:function:: void  glm_mat3_identity_array(mat3 * __restrict mat, size_t count)
+
+    make given matrix array's each element identity matrix
+
+    Parameters:
+      | *[in,out]* **mat**  matrix array (must be aligned (16/32) if alignment is not disabled)
+      | *[in]* **count**  count of matrices
+
+.. c:function:: void  glm_mat3_zero(mat3 mat)
+
+    make given matrix zero
+
+    Parameters:
+      | *[in,out]* **mat**  matrix to
+
+.. c:function:: void  glm_mat3_mul(mat3 m1, mat3 m2, mat3 dest)
+
+    multiply m1 and m2 to dest
+    m1, m2 and dest matrices can be same matrix, it is possible to write this:
+
+    .. code-block:: c
+
+       mat3 m = GLM_MAT3_IDENTITY_INIT;
+       glm_mat3_mul(m, m, m);
+
+    Parameters:
+      | *[in]*  **m1**    left matrix
+      | *[in]*  **m2**    right matrix
+      | *[out]* **dest**  destination matrix
+
+.. c:function:: void  glm_mat3_transpose_to(mat3 m, mat3 dest)
+
+    transpose mat4 and store in dest
+    source matrix will not be transposed unless dest is m
+
+    Parameters:
+      | *[in]*  **mat**   source
+      | *[out]* **dest**  destination
+
+.. c:function:: void  glm_mat3_transpose(mat3 m)
+
+    tranpose mat3 and store result in same matrix
+
+    Parameters:
+      | *[in]*  **mat**   source
+      | *[out]* **dest**  destination
+
+.. c:function:: void  glm_mat3_mulv(mat3 m, vec3 v, vec3 dest)
+
+    multiply mat4 with vec4 (column vector) and store in dest vector
+
+    Parameters:
+      | *[in]*  **mat**   mat3 (left)
+      | *[in]*  **v**     vec3 (right, column vector)
+      | *[out]* **dest**  destination (result, column vector)
+
+.. c:function:: void  glm_mat3_quat(mat3 m, versor dest)
+
+    convert mat3 to quaternion
+
+    Parameters:
+      | *[in]*  **m**     rotation matrix
+      | *[out]* **dest**  destination quaternion
+
+.. c:function:: void  glm_mat3_scale(mat3 m, float s)
+
+    multiply matrix with scalar
+
+    Parameters:
+      | *[in, out]* **mat**   matrix
+      | *[in]*      **dest**  scalar
+
+.. c:function:: float  glm_mat3_det(mat3 mat)
+
+    returns mat3 determinant
+
+    Parameters:
+      | *[in]*  **mat**   matrix
+
+    Returns:
+        mat3 determinant
+
+.. c:function:: void glm_mat3_inv(mat3 mat, mat3 dest)
+
+    inverse mat3 and store in dest
+
+    Parameters:
+      | *[in]*  **mat**  matrix
+      | *[out]* **dest** destination (inverse matrix)
+
+.. c:function:: void glm_mat3_trace(mat3 m)
+
+    | sum of the elements on the main diagonal from upper left to the lower right
+
+    Parameters:
+      | *[in]*  **m**  matrix
+
+    Returns:
+        trace of matrix
+
+.. c:function:: void  glm_mat3_swap_col(mat3 mat, int col1, int col2)
+
+    swap two matrix columns
+
+    Parameters:
+      | *[in, out]*  **mat**   matrix
+      | *[in]*       **col1**  col1
+      | *[in]*       **col2**  col2
+
+.. c:function:: void  glm_mat3_swap_row(mat3 mat, int row1, int row2)
+
+    swap two matrix rows
+
+    Parameters:
+      | *[in, out]*  **mat**   matrix
+      | *[in]*       **row1**  row1
+      | *[in]*       **row2**  row2
+
+.. c:function:: float  glm_mat3_rmc(vec3 r, mat3 m, vec3 c)
+
+    | **rmc** stands for **Row** * **Matrix** * **Column**
+
+    | helper for  R (row vector) * M (matrix) * C (column vector)
+
+    | the result is scalar because R * M = Matrix1x3 (row vector),
+    | then Matrix1x3 * Vec3 (column vector) = Matrix1x1 (Scalar)
+
+    Parameters:
+      | *[in]*  **r**  row vector or matrix1x3
+      | *[in]*  **m**  matrix3x3
+      | *[in]*  **c**  column vector or matrix3x1
+
+    Returns:
+        scalar value e.g. Matrix1x1

docs/source/mat4.rst

@@ -0,0 +1,298 @@
+.. default-domain:: C
+
+mat4
+====
+
+Header: cglm/mat4.h
+
+Important: :c:func:`glm_mat4_scale` multiplies mat4 with scalar, if you need to
+apply scale transform use :c:func:`glm_scale` functions.
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Macros:
+
+1. GLM_MAT4_IDENTITY_INIT
+#. GLM_MAT4_ZERO_INIT
+#. GLM_MAT4_IDENTITY
+#. GLM_MAT4_ZERO
+#. glm_mat4_udup(mat, dest)
+#. glm_mat4_dup(mat, dest)
+
+Functions:
+
+1. :c:func:`glm_mat4_ucopy`
+#. :c:func:`glm_mat4_copy`
+#. :c:func:`glm_mat4_identity`
+#. :c:func:`glm_mat4_identity_array`
+#. :c:func:`glm_mat4_zero`
+#. :c:func:`glm_mat4_pick3`
+#. :c:func:`glm_mat4_pick3t`
+#. :c:func:`glm_mat4_ins3`
+#. :c:func:`glm_mat4_mul`
+#. :c:func:`glm_mat4_mulN`
+#. :c:func:`glm_mat4_mulv`
+#. :c:func:`glm_mat4_mulv3`
+#. :c:func:`glm_mat3_trace`
+#. :c:func:`glm_mat3_trace3`
+#. :c:func:`glm_mat4_quat`
+#. :c:func:`glm_mat4_transpose_to`
+#. :c:func:`glm_mat4_transpose`
+#. :c:func:`glm_mat4_scale_p`
+#. :c:func:`glm_mat4_scale`
+#. :c:func:`glm_mat4_det`
+#. :c:func:`glm_mat4_inv`
+#. :c:func:`glm_mat4_inv_fast`
+#. :c:func:`glm_mat4_swap_col`
+#. :c:func:`glm_mat4_swap_row`
+#. :c:func:`glm_mat4_rmc`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: void  glm_mat4_ucopy(mat4 mat, mat4 dest)
+
+    copy mat4 to another one (dest). u means align is not required for dest
+
+    Parameters:
+      | *[in]*  **mat**   source
+      | *[out]* **dest**  destination
+
+.. c:function:: void  glm_mat4_copy(mat4 mat, mat4 dest)
+
+    copy mat4 to another one (dest).
+
+    Parameters:
+      | *[in]*  **mat**   source
+      | *[out]* **dest**  destination
+
+.. c:function:: void  glm_mat4_identity(mat4 mat)
+
+    copy identity mat4 to mat, or makes mat to identiy
+
+    Parameters:
+      | *[out]* **mat**  matrix
+
+.. c:function:: void  glm_mat4_identity_array(mat4 * __restrict mat, size_t count)
+
+    make given matrix array's each element identity matrix
+
+    Parameters:
+      | *[in,out]* **mat**  matrix array (must be aligned (16/32) if alignment is not disabled)
+      | *[in]* **count**  count of matrices
+
+.. c:function:: void  glm_mat4_zero(mat4 mat)
+
+    make given matrix zero
+
+    Parameters:
+      | *[in,out]* **mat**  matrix to
+
+.. c:function:: void  glm_mat4_pick3(mat4 mat, mat3 dest)
+
+    copy upper-left of mat4 to mat3
+
+    Parameters:
+      | *[in]*  **mat**   source
+      | *[out]* **dest**  destination
+
+.. c:function:: void  glm_mat4_pick3t(mat4 mat, mat4 dest)
+
+    copy upper-left of mat4 to mat3 (transposed)
+    the postfix t stands for transpose
+
+    Parameters:
+      | *[in]*  **mat**   source
+      | *[out]* **dest**  destination
+
+.. c:function:: void  glm_mat4_ins3(mat3 mat, mat4 dest)
+
+    copy mat3 to mat4's upper-left. this function does not fill mat4's other
+    elements. To do that use glm_mat4.
+
+    Parameters:
+      | *[in]*  **mat**   source
+      | *[out]* **dest**  destination
+
+.. c:function:: void  glm_mat4_mul(mat4 m1, mat4 m2, mat4 dest)
+
+    multiply m1 and m2 to dest
+    m1, m2 and dest matrices can be same matrix, it is possible to write this:
+
+    .. code-block:: c
+
+       mat4 m = GLM_MAT4_IDENTITY_INIT;
+       glm_mat4_mul(m, m, m);
+
+    Parameters:
+      | *[in]*  **m1**    left matrix
+      | *[in]*  **m2**    right matrix
+      | *[out]* **dest**  destination matrix
+
+.. c:function:: void glm_mat4_mulN(mat4 * __restrict matrices[], int len, mat4 dest)
+
+    mupliply N mat4 matrices and store result in dest
+    | this function lets you multiply multiple (more than two or more...)
+    | matrices
+
+    | multiplication will be done in loop, this may reduce instructions
+    | size but if **len** is too small then compiler may unroll whole loop
+
+    .. code-block:: c
+
+       mat m1, m2, m3, m4, res;
+       glm_mat4_mulN((mat4 *[]){&m1, &m2, &m3, &m4}, 4, res);
+
+    Parameters:
+      | *[in]*  **matrices** array of mat4
+      | *[in]*  **len**      matrices count
+      | *[out]* **dest**     destination matrix
+
+.. c:function:: void  glm_mat4_mulv(mat4 m, vec4 v, vec4 dest)
+
+    multiply mat4 with vec4 (column vector) and store in dest vector
+
+    Parameters:
+      | *[in]*  **m**     mat4 (left)
+      | *[in]*  **v**     vec4 (right, column vector)
+      | *[out]* **dest**  vec4 (result, column vector)
+
+.. c:function:: void  glm_mat4_mulv3(mat4 m, vec3 v, vec3 dest)
+
+    multiply vector with mat4's mat3 part(rotation)
+
+    Parameters:
+    | *[in]*  **m**     mat4 (left)
+    | *[in]*  **v**     vec3 (right, column vector)
+    | *[out]* **dest**  vec3 (result, column vector)
+
+.. c:function:: void  glm_mat4_trace(mat4 m)
+
+    | sum of the elements on the main diagonal from upper left to the lower right
+
+    Parameters:
+      | *[in]*  **m**  matrix
+
+    Returns:
+        trace of matrix
+
+.. c:function:: void  glm_mat4_trace3(mat4 m)
+
+    | trace of matrix (rotation part)
+    | sum of the elements on the main diagonal from upper left to the lower right
+
+    Parameters:
+      | *[in]*  **m**  matrix
+
+    Returns:
+        trace of matrix
+
+.. c:function:: void  glm_mat4_quat(mat4 m, versor dest)
+
+    convert mat4's rotation part to quaternion
+
+    Parameters:
+    | *[in]*  **m**     affine matrix
+    | *[out]* **dest**  destination quaternion
+
+.. c:function:: void  glm_mat4_transpose_to(mat4 m, mat4 dest)
+
+    transpose mat4 and store in dest
+    source matrix will not be transposed unless dest is m
+
+    Parameters:
+      | *[in]*  **m**     matrix
+      | *[out]* **dest**  destination matrix
+
+.. c:function:: void  glm_mat4_transpose(mat4 m)
+
+    tranpose mat4 and store result in same matrix
+
+    Parameters:
+      | *[in]*  **m**     source
+      | *[out]* **dest**  destination matrix
+
+.. c:function:: void  glm_mat4_scale_p(mat4 m, float s)
+
+    scale (multiply with scalar) matrix without simd optimization
+
+    Parameters:
+      | *[in, out]*  **m**  matrix
+      | *[in]*       **s**  scalar
+
+.. c:function:: void  glm_mat4_scale(mat4 m, float s)
+
+    scale (multiply with scalar) matrix
+    THIS IS NOT SCALE TRANSFORM, use glm_scale for that.
+
+    Parameters:
+      | *[in, out]*  **m**  matrix
+      | *[in]*       **s**  scalar
+
+.. c:function:: float  glm_mat4_det(mat4 mat)
+
+    mat4 determinant
+
+    Parameters:
+      | *[in]*  **mat**   matrix
+
+    Return:
+      | determinant
+
+.. c:function:: void  glm_mat4_inv(mat4 mat, mat4 dest)
+
+    inverse mat4 and store in dest
+
+    Parameters:
+      | *[in]*  **mat**   source
+      | *[out]* **dest**  destination matrix (inverse matrix)
+
+.. c:function:: void  glm_mat4_inv_fast(mat4 mat, mat4 dest)
+
+    inverse mat4 and store in dest
+
+    | this func uses reciprocal approximation without extra corrections
+    | e.g Newton-Raphson. this should work faster than normal,
+    | to get more precise use glm_mat4_inv version.
+
+    | NOTE: You will lose precision, glm_mat4_inv is more accurate
+
+    Parameters:
+      | *[in]*  **mat**   source
+      | *[out]* **dest**  destination
+
+.. c:function:: void  glm_mat4_swap_col(mat4 mat, int col1, int col2)
+
+    swap two matrix columns
+
+    Parameters:
+      | *[in, out]*  **mat**   matrix
+      | *[in]*       **col1**  col1
+      | *[in]*       **col2**  col2
+
+.. c:function:: void  glm_mat4_swap_row(mat4 mat, int row1, int row2)
+
+    swap two matrix rows
+
+    Parameters:
+      | *[in, out]*  **mat**   matrix
+      | *[in]*       **row1**  row1
+      | *[in]*       **row2**  row2
+
+.. c:function:: float  glm_mat4_rmc(vec4 r, mat4 m, vec4 c)
+
+    | **rmc** stands for **Row** * **Matrix** * **Column**
+
+    | helper for  R (row vector) * M (matrix) * C (column vector)
+
+    | the result is scalar because R * M = Matrix1x4 (row vector),
+    | then Matrix1x4 * Vec4 (column vector) = Matrix1x1 (Scalar)
+
+    Parameters:
+      | *[in]*  **r**  row vector or matrix1x4
+      | *[in]*  **m**  matrix4x4
+      | *[in]*  **c**  column vector or matrix4x1
+
+    Returns:
+        scalar value e.g. Matrix1x1

docs/source/opengl.rst

@@ -0,0 +1,61 @@
+How to send vector or matrix to OpenGL like API
+==================================================
+
+*cglm*'s vector and matrix types are arrays. So you can send them directly to a
+function which accecpts pointer. But you may got warnings for matrix because it is
+two dimensional array.
+
+Passing / Uniforming Matrix to OpenGL:
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+**glUniformMatrix4fv** accepts float pointer, you can pass matrix to that parameter
+and it should work but with warnings. "You can pass" doesn't mean that you must pass like that.
+
+**Correct options:**
+
+Correct doesn't mean correct way to use OpenGL it is just shows correct way to pass cglm type to it.
+
+1. Pass first column
+---------------------
+
+The goal is that pass address of matrix, first element of matrix is also address of matrix,
+because it is array of vectors and vector is array of floats.
+
+.. code-block:: c
+
+   mat4 matrix;
+   /* ... */
+   glUniformMatrix4fv(location, 1, GL_FALSE, matrix[0]);
+
+array of matrices:
+
+.. code-block:: c
+
+   mat4 matrix;
+   /* ... */
+   glUniformMatrix4fv(location, count, GL_FALSE, matrix[0][0]);
+
+1. Cast matrix to pointer
+--------------------------
+
+.. code-block:: c
+
+   mat4 matrix;
+   /* ... */
+   glUniformMatrix4fv(location, count, GL_FALSE, (float *)matrix);
+
+in this way, passing aray of matrices is same
+
+Passing / Uniforming Vectors to OpenGL:
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+You don't need to do extra thing when passing cglm vectors to OpengL or other APIs.
+Because a function like **glUniform4fv** accepts vector as pointer. cglm's vectors
+are array of floats. So you can pass it directly ot those functions:
+
+.. code-block:: c
+
+   vec4 vec;
+   /* ... */
+   glUniform4fv(location, 1, vec);
+
+this show how to pass **vec4** others are same.

docs/source/opt.rst

@@ -0,0 +1,52 @@
+.. default-domain:: C
+
+Options
+===============================================================================
+
+A few options are provided via macros.
+
+Alignment Option
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+As default, cglm requires types to be aligned. Alignment requirements:
+
+vec3:   8 byte
+vec4:   16 byte
+mat4:   16 byte
+versor: 16 byte
+
+By starting **v0.4.5** cglm provides an option to disable alignment requirement.
+To enable this option define **CGLM_ALL_UNALIGNED** macro before all headers.
+You can define it in Xcode, Visual Studio (or other IDEs) or you can also prefer
+to define it in build system. If you use pre-compiled verisons then you
+have to compile cglm with **CGLM_ALL_UNALIGNED** macro.
+
+**VERY VERY IMPORTANT:** If you use cglm in multiple projects and
+ those projects are depends on each other, then
+
+ | *ALWAYS* or *NEVER USE* **CGLM_ALL_UNALIGNED** macro in linked projects
+
+ if you do not know what you are doing. Because a cglm header included
+ via 'project A' may force types to be aligned and another cglm header
+ included via 'project B' may not require alignment. In this case
+ cglm functions will read from and write to **INVALID MEMORY LOCATIONs**.
+
+ ALWAYS USE SAME CONFIGURATION / OPTION for **cglm** if you have multiple projects.
+
+ For instance if you set CGLM_ALL_UNALIGNED in a project then set it in other projects too
+
+SSE and SSE2 Shuffle Option
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+**_mm_shuffle_ps** generates **shufps** instruction even if registers are same.
+You can force it to generate **pshufd** instruction by defining
+**CGLM_USE_INT_DOMAIN** macro. As default it is not defined.
+
+SSE3 and SSE4 Dot Product Options
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+You have to extra options for dot product: **CGLM_SSE4_DOT** and **CGLM_SSE3_DOT**.
+
+- If **SSE4** is enabled then you can define **CGLM_SSE4_DOT** to force cglm to use **_mm_dp_ps** instruction.
+- If **SSE3** is enabled then you can define **CGLM_SSE3_DOT** to force cglm to use **_mm_hadd_ps** instructions.
+
+otherwise cglm will use custom cglm's hadd functions which are optimized too.

docs/source/plane.rst

@@ -0,0 +1,33 @@
+.. default-domain:: C
+
+plane
+================================================================================
+
+Header: cglm/plane.h
+
+Plane extract functions are in frustum header and documented
+in :doc:`frustum` page.
+
+**Definition of plane:**
+
+Plane equation:  **Ax + By + Cz + D = 0**
+
+Plan is stored in **vec4** as **[A, B, C, D]**. (A, B, C) is normal and D is distance
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Functions:
+
+1. :c:func:`glm_plane_normalize`
+
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: void  glm_plane_normalize(vec4 plane)
+
+    | normalizes a plane
+
+    Parameters:
+      | *[in, out]*  **plane**  pnale to normalize

docs/source/project.rst

@@ -0,0 +1,102 @@
+.. default-domain:: C
+
+Project / UnProject
+================================================================================
+
+Header: cglm/project.h
+
+Viewport is required as *vec4* **[X, Y, Width, Height]** but this doesn't mean
+that you should store it as **vec4**. You can convert your data representation
+to vec4 before passing it to related functions.
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Functions:
+
+1. :c:func:`glm_unprojecti`
+#. :c:func:`glm_unproject`
+#. :c:func:`glm_project`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: void  glm_unprojecti(vec3 pos, mat4 invMat, vec4 vp, vec3 dest)
+
+    | maps the specified viewport coordinates into specified space [1]
+      the matrix should contain projection matrix.
+
+    if you don't have ( and don't want to have ) an inverse matrix then use
+    glm_unproject version. You may use existing inverse of matrix in somewhere
+    else, this is why glm_unprojecti exists to save save inversion cost
+
+    [1] space:
+      - if m = invProj:     View Space
+      - if m = invViewProj: World Space
+      - if m = invMVP:      Object Space
+
+    You probably want to map the coordinates into object space
+    so use invMVP as m
+
+    Computing viewProj:
+
+    .. code-block:: c
+
+       glm_mat4_mul(proj, view, viewProj);
+       glm_mat4_mul(viewProj, model, MVP);
+       glm_mat4_inv(viewProj, invMVP);
+
+    Parameters:
+      | *[in]*  **pos**     point/position in viewport coordinates
+      | *[in]*  **invMat**  matrix (see brief)
+      | *[in]*  **vp**      viewport as [x, y, width, height]
+      | *[out]* **dest**    unprojected coordinates
+
+.. c:function:: void  glm_unproject(vec3 pos, mat4 m, vec4 vp, vec3 dest)
+
+    | maps the specified viewport coordinates into specified space [1]
+      the matrix should contain projection matrix.
+
+    this is same as glm_unprojecti except this function get inverse matrix for
+    you.
+
+    [1] space:
+      - if m = proj:     View Space
+      - if m = viewProj: World Space
+      - if m = MVP:      Object Space
+
+    You probably want to map the coordinates into object space so use MVP as m
+
+    Computing viewProj and MVP:
+
+    .. code-block:: c
+
+       glm_mat4_mul(proj, view, viewProj);
+       glm_mat4_mul(viewProj, model, MVP);
+
+    Parameters:
+      | *[in]*  **pos**  point/position in viewport coordinates
+      | *[in]*  **m**    matrix (see brief)
+      | *[in]*  **vp**   viewport as [x, y, width, height]
+      | *[out]* **dest** unprojected coordinates
+
+.. c:function:: void  glm_project(vec3 pos, mat4 m, vec4 vp, vec3 dest)
+
+    | map object coordinates to window coordinates
+
+    Computing MVP:
+
+    .. code-block:: c
+
+       glm_mat4_mul(proj, view, viewProj);
+       glm_mat4_mul(viewProj, model, MVP);
+
+    this could be useful for gettng a bbox which fits with view frustum and
+    object bounding boxes. In this case you crop view frustum box with objects
+    box
+
+    Parameters:
+      | *[in]*  **pos**      object coordinates
+      | *[in]*  **m**        MVP matrix
+      | *[in]*  **vp**       viewport as [x, y, width, height]
+      | *[out]* **dest**     projected coordinates

docs/source/quat.rst

@@ -0,0 +1,387 @@
+.. default-domain:: C
+
+quaternions
+===========
+
+Header: cglm/quat.h
+
+ **Important:** *cglm* stores quaternion as **[x, y, z, w]** in memory
+ since **v0.4.0** it was **[w, x, y, z]**
+ before v0.4.0 ( **v0.3.5 and earlier** ). w is real part.
+
+What you can do with quaternions with existing functions is (Some of them):
+
+- You can rotate transform matrix using quaterion
+- You can rotate vector using quaterion
+- You can create view matrix using quaterion
+- You can create a lookrotation (from source point to dest)
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Macros:
+
+1. GLM_QUAT_IDENTITY_INIT
+#. GLM_QUAT_IDENTITY
+
+Functions:
+
+1. :c:func:`glm_quat_identity`
+#. :c:func:`glm_quat_identity_array`
+#. :c:func:`glm_quat_init`
+#. :c:func:`glm_quat`
+#. :c:func:`glm_quatv`
+#. :c:func:`glm_quat_copy`
+#. :c:func:`glm_quat_norm`
+#. :c:func:`glm_quat_normalize`
+#. :c:func:`glm_quat_normalize_to`
+#. :c:func:`glm_quat_dot`
+#. :c:func:`glm_quat_conjugate`
+#. :c:func:`glm_quat_inv`
+#. :c:func:`glm_quat_add`
+#. :c:func:`glm_quat_sub`
+#. :c:func:`glm_quat_real`
+#. :c:func:`glm_quat_imag`
+#. :c:func:`glm_quat_imagn`
+#. :c:func:`glm_quat_imaglen`
+#. :c:func:`glm_quat_angle`
+#. :c:func:`glm_quat_axis`
+#. :c:func:`glm_quat_mul`
+#. :c:func:`glm_quat_mat4`
+#. :c:func:`glm_quat_mat4t`
+#. :c:func:`glm_quat_mat3`
+#. :c:func:`glm_quat_mat3t`
+#. :c:func:`glm_quat_lerp`
+#. :c:func:`glm_quat_slerp`
+#. :c:func:`glm_quat_look`
+#. :c:func:`glm_quat_for`
+#. :c:func:`glm_quat_forp`
+#. :c:func:`glm_quat_rotatev`
+#. :c:func:`glm_quat_rotate`
+#. :c:func:`glm_quat_rotate_at`
+#. :c:func:`glm_quat_rotate_atm`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: void  glm_quat_identity(versor q)
+
+    | makes given quat to identity
+
+    Parameters:
+      | *[in, out]*  **q**    quaternion
+
+.. c:function:: void  glm_quat_identity_array(versor * __restrict q, size_t count)
+
+    | make given quaternion array's each element identity quaternion
+
+    Parameters:
+      | *[in, out]*  **q**   quat array (must be aligned (16) if alignment is not disabled)
+      | *[in]*  **count**    count of quaternions
+
+.. c:function:: void  glm_quat_init(versor q, float x, float y, float z, float w)
+
+    | inits quaternion with given values
+
+    Parameters:
+      | *[out]* **q**      quaternion
+      | *[in]*  **x**      imag.x
+      | *[in]*  **y**      imag.y
+      | *[in]*  **z**      imag.z
+      | *[in]*  **w**      w (real part)
+
+.. c:function:: void  glm_quat(versor q, float  angle, float  x, float  y, float  z)
+
+    | creates NEW quaternion with individual axis components
+
+    | given axis will be normalized
+
+    Parameters:
+      | *[out]* **q**      quaternion
+      | *[in]*  **angle**  angle (radians)
+      | *[in]*  **x**      axis.x
+      | *[in]*  **y**      axis.y
+      | *[in]*  **z**      axis.z
+
+.. c:function:: void  glm_quatv(versor q, float  angle, vec3   axis)
+
+    | creates NEW quaternion with axis vector
+
+    | given axis will be normalized
+
+    Parameters:
+      | *[out]* **q**      quaternion
+      | *[in]*  **angle**  angle (radians)
+      | *[in]*  **axis**   axis (will be normalized)
+
+.. c:function:: void  glm_quat_copy(versor q, versor dest)
+
+    | copy quaternion to another one
+
+    Parameters:
+      | *[in]*  **q**     source quaternion
+      | *[out]* **dest**  destination quaternion
+
+.. c:function:: float  glm_quat_norm(versor q)
+
+    | returns norm (magnitude) of quaternion
+
+    Parameters:
+      | *[in]*  **a**  quaternion
+
+    Returns:
+      norm (magnitude)
+
+.. c:function:: void  glm_quat_normalize_to(versor q, versor dest)
+
+    | normalize quaternion and store result in dest, original one will not be normalized
+
+    Parameters:
+      | *[in]*  **q**    quaternion to normalize into
+      | *[out]* **dest** destination quaternion
+
+.. c:function:: void  glm_quat_normalize(versor q)
+
+    | normalize quaternion
+
+    Parameters:
+      | *[in, out]*  **q** quaternion
+
+.. c:function:: float  glm_quat_dot(versor p, versor q)
+
+    dot product of two quaternion
+
+    Parameters:
+      | *[in]*  **p**   quaternion 1
+      | *[in]*  **q**   quaternion 2
+
+    Returns:
+      dot product
+
+.. c:function:: void  glm_quat_conjugate(versor q, versor dest)
+
+    conjugate of quaternion
+
+    Parameters:
+      | *[in]*  **q**      quaternion
+      | *[in]*  **dest**   conjugate
+
+.. c:function:: void  glm_quat_inv(versor q, versor dest)
+
+    inverse of non-zero quaternion
+
+    Parameters:
+      | *[in]*  **q**      quaternion
+      | *[in]*  **dest**   inverse quaternion
+
+.. c:function:: void  glm_quat_add(versor p, versor q, versor dest)
+
+    add (componentwise) two quaternions and store result in dest
+
+    Parameters:
+      | *[in]*  **p**      quaternion 1
+      | *[in]*  **q**      quaternion 2
+      | *[in]*  **dest**   result quaternion
+
+.. c:function:: void  glm_quat_sub(versor p, versor q, versor dest)
+
+    subtract (componentwise) two quaternions and store result in dest
+
+    Parameters:
+      | *[in]*  **p**      quaternion 1
+      | *[in]*  **q**      quaternion 2
+      | *[in]*  **dest**   result quaternion
+
+.. c:function:: float  glm_quat_real(versor q)
+
+    returns real part of quaternion
+
+    Parameters:
+      | *[in]*  **q**   quaternion
+
+    Returns:
+      real part (quat.w)
+
+.. c:function:: void  glm_quat_imag(versor q, vec3 dest)
+
+    returns imaginary part of quaternion
+
+    Parameters:
+      | *[in]*   **q**      quaternion
+      | *[out]*  **dest**   imag
+
+.. c:function:: void  glm_quat_imagn(versor q, vec3 dest)
+
+    returns normalized imaginary part of quaternion
+
+    Parameters:
+      | *[in]*   **q**      quaternion
+      | *[out]*  **dest**   imag
+
+.. c:function:: float  glm_quat_imaglen(versor q)
+
+    returns length of imaginary part of quaternion
+
+    Parameters:
+      | *[in]*   **q**      quaternion
+
+    Returns:
+      norm of imaginary part
+
+.. c:function:: float  glm_quat_angle(versor q)
+
+    returns angle of quaternion
+
+    Parameters:
+      | *[in]*  **q**   quaternion
+
+    Returns:
+      angles of quat (radians)
+
+.. c:function:: void  glm_quat_axis(versor q, versor dest)
+
+    axis of quaternion
+
+    Parameters:
+      | *[in]*   **p**      quaternion
+      | *[out]*  **dest**   axis of quaternion
+
+.. c:function:: void  glm_quat_mul(versor p, versor q, versor dest)
+
+    | multiplies two quaternion and stores result in dest
+
+    | this is also called Hamilton Product
+
+    | According to WikiPedia:
+    | The product of two rotation quaternions [clarification needed] will be
+      equivalent to the rotation q followed by the rotation p
+
+    Parameters:
+      | *[in]*  **p**     quaternion 1 (first rotation)
+      | *[in]*  **q**     quaternion 2 (second rotation)
+      | *[out]* **dest**  result quaternion
+
+.. c:function:: void  glm_quat_mat4(versor q, mat4 dest)
+
+    | convert quaternion to mat4
+
+    Parameters:
+      | *[in]*  **q**     quaternion
+      | *[out]* **dest**  result matrix
+
+.. c:function:: void  glm_quat_mat4t(versor q, mat4 dest)
+
+    | convert quaternion to mat4 (transposed). This is transposed version of glm_quat_mat4
+
+    Parameters:
+      | *[in]*  **q**     quaternion
+      | *[out]* **dest**  result matrix
+
+.. c:function:: void  glm_quat_mat3(versor q, mat3 dest)
+
+    | convert quaternion to mat3
+
+    Parameters:
+      | *[in]*  **q**     quaternion
+      | *[out]* **dest**  result matrix
+
+.. c:function:: void  glm_quat_mat3t(versor q, mat3 dest)
+
+    | convert quaternion to mat3 (transposed). This is transposed version of glm_quat_mat3
+
+    Parameters:
+      | *[in]*  **q**     quaternion
+      | *[out]* **dest**  result matrix
+
+.. c:function:: void  glm_quat_lerp(versor from, versor to, float t, versor dest)
+
+    | interpolates between two quaternions
+    | using spherical linear interpolation (LERP)
+
+    Parameters:
+      | *[in]*  **from**  from
+      | *[in]*  **to**    to
+      | *[in]*  **t**     interpolant (amount) clamped between 0 and 1
+      | *[out]* **dest**  result quaternion
+
+.. c:function:: void glm_quat_slerp(versor q, versor r, float  t, versor dest)
+
+    | interpolates between two quaternions
+    | using spherical linear interpolation (SLERP)
+
+    Parameters:
+      | *[in]*  **from**  from
+      | *[in]*  **to**    to
+      | *[in]*  **t**     interpolant (amount) clamped between 0 and 1
+      | *[out]* **dest**  result quaternion
+
+.. c:function:: void  glm_quat_look(vec3 eye, versor ori, mat4 dest)
+
+    | creates view matrix using quaternion as camera orientation
+
+    Parameters:
+      | *[in]*  **eye**   eye
+      | *[in]*  **ori**   orientation in world space as quaternion
+      | *[out]* **dest**  result matrix
+
+.. c:function:: void  glm_quat_for(vec3 dir, vec3 up, versor dest)
+
+    | creates look rotation quaternion
+
+    Parameters:
+      | *[in]*  **dir**   direction to look
+      | *[in]*  **up**    up vector
+      | *[out]* **dest**  result matrix
+
+.. c:function:: void  glm_quat_forp(vec3 from, vec3 to, vec3 up, versor dest)
+
+    | creates look rotation quaternion using source and destination positions p suffix stands for position
+
+    | this is similar to glm_quat_for except this computes direction for glm_quat_for for you.
+
+    Parameters:
+      | *[in]*  **from**  source point
+      | *[in]*  **to**    destination point
+      | *[in]*  **up**    up vector
+      | *[out]* **dest**  result matrix
+
+.. c:function:: void  glm_quat_rotatev(versor q, vec3 v, vec3 dest)
+
+    | crotate vector using using quaternion
+
+    Parameters:
+      | *[in]*  **q**     quaternion
+      | *[in]*  **v**     vector to rotate
+      | *[out]* **dest**  rotated vector
+
+.. c:function:: void glm_quat_rotate(mat4 m, versor q, mat4 dest)
+
+    | rotate existing transform matrix using quaternion
+
+    instead of passing identity matrix, consider to use quat_mat4 functions
+
+    Parameters:
+      | *[in]*  **m**     existing transform matrix to rotate
+      | *[in]*  **q**     quaternion
+      | *[out]* **dest**  rotated matrix/transform
+
+.. c:function:: void glm_quat_rotate_at(mat4 m, versor q, vec3 pivot)
+
+    | rotate existing transform matrix using quaternion at pivot point
+
+    Parameters:
+      | *[in, out]*  **m**      existing transform matrix to rotate
+      | *[in]*       **q**      quaternion
+      | *[in]*       **pivot**  pivot
+
+.. c:function:: void glm_quat_rotate(mat4 m, versor q, mat4 dest)
+
+    | rotate NEW transform matrix using quaternion at pivot point
+    | this creates rotation matrix, it assumes you don't have a matrix
+
+    | this should work faster than glm_quat_rotate_at because it reduces one glm_translate.
+
+    Parameters:
+      | *[in, out]*  **m**      existing transform matrix to rotate
+      | *[in]*       **q**      quaternion
+      | *[in]*       **pivot**  pivot

docs/source/sphere.rst

@@ -0,0 +1,74 @@
+.. default-domain:: C
+
+Sphere
+================================================================================
+
+Header: cglm/sphere.h
+
+**Definition of sphere:**
+
+Sphere Representation in cglm is *vec4*: **[center.x, center.y, center.z, radii]**
+
+You can call any vec3 function by pasing sphere. Because first three elements
+defines center of sphere.
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Functions:
+
+1. :c:func:`glm_sphere_radii`
+#. :c:func:`glm_sphere_transform`
+#. :c:func:`glm_sphere_merge`
+#. :c:func:`glm_sphere_sphere`
+#. :c:func:`glm_sphere_point`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: float  glm_sphere_radii(vec4 s)
+
+    | helper for getting sphere radius
+
+    Parameters:
+      | *[in]*  **s**   sphere
+
+    Returns:
+       returns radii
+
+.. c:function:: void  glm_sphere_transform(vec4 s, mat4 m, vec4 dest)
+
+    | apply transform to sphere, it is just wrapper for glm_mat4_mulv3
+
+    Parameters:
+      | *[in]*  **s**    sphere
+      | *[in]*  **m**    transform matrix
+      | *[out]* **dest** transformed sphere
+
+.. c:function:: void  glm_sphere_merge(vec4 s1, vec4 s2, vec4 dest)
+
+    | merges two spheres and creates a new one
+
+    two sphere must be in same space, for instance if one in world space then
+    the other must be in world space too, not in local space.
+
+    Parameters:
+      | *[in]*  **s1**      sphere 1
+      | *[in]*  **s2**      sphere 2
+      | *[out]* **dest**    merged/extended sphere
+
+.. c:function:: bool  glm_sphere_sphere(vec4 s1, vec4 s2)
+
+    | check if two sphere intersects
+
+    Parameters:
+      | *[in]*  **s1**      sphere
+      | *[in]*  **s2**      other sphere
+
+.. c:function:: bool  glm_sphere_point(vec4 s, vec3 point)
+
+    | check if sphere intersects with point
+
+    Parameters:
+      | *[in]*  **s**       sphere
+      | *[in]*  **point**   point

docs/source/troubleshooting.rst

@@ -0,0 +1,86 @@
+.. default-domain:: C
+
+Troubleshooting
+================================================================================
+
+It is possible that sometimes you may get crashes or wrong results.
+Follow these topics
+
+Memory Allocation:
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Again, **cglm** doesn't alloc any memory on heap.
+cglm functions works like memcpy; it copies data from src,
+makes calculations then copy the result to dest.
+
+You are responsible for allocation of **src** and **dest** parameters.
+
+Alignment:
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+**vec4** and **mat4** types requires 16 byte alignment.
+These types are marked with align attribute to let compiler know about this
+requirement.
+
+But since MSVC (Windows) throws the error:
+
+**"formal parameter with requested alignment of 16 won't be aligned"**
+
+The alignment attribute has been commented for MSVC
+
+.. code-block:: c
+
+   #if defined(_MSC_VER)
+   #  define CGLM_ALIGN(X) /* __declspec(align(X)) */
+   #else
+   #  define CGLM_ALIGN(X) __attribute((aligned(X)))
+   #endif.
+
+So MSVC may not know about alignment requirements when creating variables.
+The interesting thing is that, if I remember correctly Visual Studio 2017
+doesn't throw the above error. So we may uncomment that line for Visual Studio 2017,
+you may do it yourself.
+
+**This MSVC issue is still in TODOs.**
+
+**UPDATE:** By starting v0.4.5 cglm provides an option to disable alignment requirement.
+Also alignment is disabled for older msvc verisons as default. Now alignment is only required in Visual Studio 2017 version 15.6+ if CGLM_ALL_UNALIGNED macro is not defined.
+
+Crashes, Invalid Memory Access:
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Probably you are trying to write to invalid memory location.
+
+You may used wrong function for what you want to do.
+
+For instance you may called **glm_vec4_** functions for **vec3** data type.
+It will try to write 32 byte but since **vec3** is 24 byte it should throw
+memory access error or exit the app without saying anything.
+
+**UPDATE - IMPORTANT:** 
+
+  | On MSVC or some other compilers, if alignment is enabled (default) then double check alignment requirements if you got a crash.
+
+  | If you send GLM_VEC4_ONE or similar macros directly to a function, it may be crashed.
+  | Because compiler may not apply alignment as defined on **typedef** to that macro while passing it (on stack) to a function.
+
+Wrong Results:
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Again, you may used wrong function.
+
+For instance if you use **glm_normalize()** or **glm_vec3_normalize()** for **vec4**,
+it will assume that passed param is **vec3** and will normalize it for **vec3**.
+Since you need to **vec4** to be normalized in your case, you will get wrong results.
+
+Accessing vec4 type with vec3 functions is valid, you will not get any error, exception or crash.
+You only get wrong results if you don't know what you are doing!
+
+So be carefull, when your IDE (Xcode, Visual Studio ...) tried to autocomplete function names, READ IT :)
+
+**Also implementation may be wrong please let us know by creating an issue on Github.**
+
+Other Issues?
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+**Please let us know by creating an issue on Github.**

docs/source/util.rst

@@ -0,0 +1,173 @@
+.. default-domain:: C
+
+utils / helpers
+================================================================================
+
+Header: cglm/util.h
+
+
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Functions:
+
+1. :c:func:`glm_sign`
+#. :c:func:`glm_signf`
+#. :c:func:`glm_rad`
+#. :c:func:`glm_deg`
+#. :c:func:`glm_make_rad`
+#. :c:func:`glm_make_deg`
+#. :c:func:`glm_pow2`
+#. :c:func:`glm_min`
+#. :c:func:`glm_max`
+#. :c:func:`glm_clamp`
+#. :c:func:`glm_lerp`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: int  glm_sign(int val)
+
+    | returns sign of 32 bit integer as +1, -1, 0
+
+    | **Important**: It returns 0 for zero input
+
+    Parameters:
+      | *[in]*  **val**   an integer
+
+    Returns:
+      sign of given number
+
+.. c:function:: float  glm_signf(float val)
+
+    | returns sign of 32 bit integer as +1.0, -1.0, 0.0
+
+    | **Important**: It returns 0.0f for zero input
+
+    Parameters:
+      | *[in]*  **val**   a float
+
+    Returns:
+      sign of given number
+
+.. c:function:: float  glm_rad(float deg)
+
+    | convert degree to radians
+
+    Parameters:
+      | *[in]*  **deg**   angle in degrees
+
+.. c:function:: float glm_deg(float rad)
+
+    | convert radians to degree
+
+    Parameters:
+      | *[in]*  **rad**      angle in radians
+
+.. c:function:: void  glm_make_rad(float *degm)
+
+    | convert exsisting degree to radians. this will override degrees value
+
+    Parameters:
+      | *[in, out]*  **deg**      pointer to angle in degrees
+
+.. c:function:: void  glm_make_deg(float *rad)
+
+    | convert exsisting radians to degree. this will override radians value
+
+    Parameters:
+      | *[in, out]*  **rad**      pointer to angle in radians
+
+.. c:function:: float  glm_pow2(float x)
+
+    | multiplies given parameter with itself = x * x or powf(x, 2)
+
+    Parameters:
+      | *[in]*  **x** value
+
+    Returns:
+      square of a given number
+
+.. c:function:: float  glm_min(float a, float b)
+
+    | returns minimum of given two values
+
+    Parameters:
+      | *[in]*  **a** number 1
+      | *[in]*  **b** number 2
+
+    Returns:
+      minimum value
+
+.. c:function:: float  glm_max(float a, float b)
+
+    | returns maximum of given two values
+
+    Parameters:
+      | *[in]*  **a** number 1
+      | *[in]*  **b** number 2
+
+    Returns:
+      maximum value
+
+.. c:function:: void  glm_clamp(float val, float minVal, float maxVal)
+
+    constrain a value to lie between two further values
+
+    Parameters:
+      | *[in]*  **val**     input value
+      | *[in]*  **minVal**  minimum value
+      | *[in]*  **maxVal**  maximum value
+
+    Returns:
+      clamped value
+
+.. c:function:: float  glm_lerp(float from, float to, float t)
+
+    linear interpolation between two number
+
+    | formula:  from + s * (to - from)
+
+    Parameters:
+      | *[in]*  **from**   from value
+      | *[in]*  **to**     to value
+      | *[in]*  **t**      interpolant (amount) clamped between 0 and 1
+
+    Returns:
+       interpolated value
+
+.. c:function:: bool glm_eq(float a, float b)
+
+    check if two float equal with using EPSILON
+
+    Parameters:
+      | *[in]*  **a**   a
+      | *[in]*  **b**   b
+
+    Returns:
+       true if a and b equals
+
+.. c:function:: float glm_percent(float from, float to, float current)
+
+    percentage of current value between start and end value
+
+    Parameters:
+      | *[in]*  **from**   from value
+      | *[in]*  **to**     to value
+      | *[in]*  **current**   value between from and to values
+
+    Returns:
+       clamped normalized percent (0-100 in 0-1)
+
+.. c:function:: float glm_percentc(float from, float to, float current)
+
+    clamped percentage of current value between start and end value
+
+    Parameters:
+      | *[in]*  **from**      from value
+      | *[in]*  **to**        to value
+      | *[in]*  **current**   value between from and to values
+
+    Returns:
+       clamped normalized percent (0-100 in 0-1)

docs/source/vec2-ext.rst

@@ -0,0 +1,134 @@
+.. default-domain:: C
+
+vec2 extra
+==========
+
+Header: cglm/vec2-ext.h
+
+There are some functions are in called in extra header. These are called extra
+because they are not used like other functions in vec2.h in the future some of
+these functions ma be moved to vec2 header.
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Functions:
+
+1. :c:func:`glm_vec2_fill`
+#. :c:func:`glm_vec2_eq`
+#. :c:func:`glm_vec2_eq_eps`
+#. :c:func:`glm_vec2_eq_all`
+#. :c:func:`glm_vec2_eqv`
+#. :c:func:`glm_vec2_eqv_eps`
+#. :c:func:`glm_vec2_max`
+#. :c:func:`glm_vec2_min`
+#. :c:func:`glm_vec2_isnan`
+#. :c:func:`glm_vec2_isinf`
+#. :c:func:`glm_vec2_isvalid`
+#. :c:func:`glm_vec2_sign`
+#. :c:func:`glm_vec2_sqrt`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: void glm_vec2_fill(vec2 v, float val)
+
+    fill a vector with specified value
+
+    Parameters:
+      | *[in,out]* **dest**  destination
+      | *[in]*     **val**   value
+      
+
+.. c:function:: bool glm_vec2_eq(vec2 v, float val)
+
+    check if vector is equal to value (without epsilon)
+
+    Parameters:
+      | *[in]*  **v**    vector
+      | *[in]*  **val**  value
+
+.. c:function:: bool glm_vec2_eq_eps(vec2 v, float val)
+
+    check if vector is equal to value (with epsilon)
+
+    Parameters:
+      | *[in]*  **v**    vector
+      | *[in]*  **val**  value
+
+.. c:function:: bool glm_vec2_eq_all(vec2 v)
+
+    check if vectors members are equal (without epsilon)
+
+    Parameters:
+      | *[in]*  **v**   vector
+
+.. c:function:: bool glm_vec2_eqv(vec2 v1, vec2 v2)
+
+    check if vector is equal to another (without epsilon) vector
+
+    Parameters:
+      | *[in]*  **vec**   vector 1
+      | *[in]*  **vec**   vector 2
+
+.. c:function:: bool glm_vec2_eqv_eps(vec2 v1, vec2 v2)
+
+    check if vector is equal to another (with epsilon)
+
+    Parameters:
+      | *[in]*  **v1**    vector1
+      | *[in]*  **v2**    vector2
+
+.. c:function:: float glm_vec2_max(vec2 v)
+
+    max value of vector
+
+    Parameters:
+      | *[in]*  **v**    vector
+
+.. c:function:: float glm_vec2_min(vec2 v)
+
+     min value of vector
+
+    Parameters:
+      | *[in]*  **v**  vector
+
+.. c:function:: bool glm_vec2_isnan(vec2 v)
+
+    | check if one of items is NaN (not a number)
+    | you should only use this in DEBUG mode or very critical asserts
+
+    Parameters:
+      | *[in]*  **v**  vector
+
+.. c:function:: bool glm_vec2_isinf(vec2 v)
+
+    | check if one of items is INFINITY
+    | you should only use this in DEBUG mode or very critical asserts
+
+    Parameters:
+      | *[in]*  **v**  vector
+
+.. c:function:: bool glm_vec2_isvalid(vec2 v)
+
+    | check if all items are valid number
+    | you should only use this in DEBUG mode or very critical asserts
+
+    Parameters:
+      | *[in]*  **v**  vector
+
+.. c:function:: void glm_vec2_sign(vec2 v, vec2 dest)
+
+    get sign of 32 bit float as +1, -1, 0
+
+    Parameters:
+      | *[in]*   **v**     vector
+      | *[out]*  **dest**  sign vector (only keeps signs as -1, 0, -1)
+
+.. c:function:: void glm_vec2_sqrt(vec2 v, vec2 dest)
+
+    square root of each vector item
+
+    Parameters:
+      | *[in]*   **v**     vector
+      | *[out]*  **dest**  destination vector (sqrt(v))

docs/source/vec2.rst

@@ -0,0 +1,375 @@
+.. default-domain:: C
+
+vec2
+====
+
+Header: cglm/vec2.h
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Macros:
+
+1. GLM_vec2_ONE_INIT
+#. GLM_vec2_ZERO_INIT
+#. GLM_vec2_ONE
+#. GLM_vec2_ZERO
+
+Functions:
+
+1. :c:func:`glm_vec2`
+#. :c:func:`glm_vec2_copy`
+#. :c:func:`glm_vec2_zero`
+#. :c:func:`glm_vec2_one`
+#. :c:func:`glm_vec2_dot`
+#. :c:func:`glm_vec2_cross`
+#. :c:func:`glm_vec2_norm2`
+#. :c:func:`glm_vec2_norm`
+#. :c:func:`glm_vec2_add`
+#. :c:func:`glm_vec2_adds`
+#. :c:func:`glm_vec2_sub`
+#. :c:func:`glm_vec2_subs`
+#. :c:func:`glm_vec2_mul`
+#. :c:func:`glm_vec2_scale`
+#. :c:func:`glm_vec2_scale_as`
+#. :c:func:`glm_vec2_div`
+#. :c:func:`glm_vec2_divs`
+#. :c:func:`glm_vec2_addadd`
+#. :c:func:`glm_vec2_subadd`
+#. :c:func:`glm_vec2_muladd`
+#. :c:func:`glm_vec2_muladds`
+#. :c:func:`glm_vec2_maxadd`
+#. :c:func:`glm_vec2_minadd`
+#. :c:func:`glm_vec2_negate`
+#. :c:func:`glm_vec2_negate_to`
+#. :c:func:`glm_vec2_normalize`
+#. :c:func:`glm_vec2_normalize_to`
+#. :c:func:`glm_vec2_rotate`
+#. :c:func:`glm_vec2_distance2`
+#. :c:func:`glm_vec2_distance`
+#. :c:func:`glm_vec2_maxv`
+#. :c:func:`glm_vec2_minv`
+#. :c:func:`glm_vec2_clamp`
+#. :c:func:`glm_vec2_lerp`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: void glm_vec2(vec4 v4, vec2 dest)
+
+    init vec2 using vec3
+
+    Parameters:
+      | *[in]*  **v3**    vector3
+      | *[out]* **dest**  destination
+
+.. c:function:: void glm_vec2_copy(vec2 a, vec2 dest)
+
+    copy all members of [a] to [dest]
+
+    Parameters:
+      | *[in]*  **a**     source
+      | *[out]* **dest**  destination
+
+.. c:function:: void glm_vec2_zero(vec2 v)
+
+    makes all members 0.0f (zero)
+
+    Parameters:
+      | *[in, out]*  **v**     vector
+
+.. c:function:: void  glm_vec2_one(vec2 v)
+
+    makes all members 1.0f (one)
+
+    Parameters:
+      | *[in, out]*  **v**     vector
+
+.. c:function:: float glm_vec2_dot(vec2 a, vec2 b)
+
+    dot product of vec2
+
+    Parameters:
+      | *[in]*  **a**  vector1
+      | *[in]*  **b**  vector2
+
+    Returns:
+      dot product
+
+.. c:function:: void glm_vec2_cross(vec2 a, vec2 b, vec2 d)
+
+    cross product of two vector (RH)
+
+    | ref: http://allenchou.net/2013/07/cross-product-of-2d-vectors/
+
+    Parameters:
+      | *[in]*  **a**     vector 1
+      | *[in]*  **b**     vector 2
+      | *[out]* **dest**  destination
+
+    Returns:
+      Z component of cross product
+
+.. c:function:: float glm_vec2_norm2(vec2 v)
+
+    norm * norm (magnitude) of vector
+
+    we can use this func instead of calling norm * norm, because it would call
+    sqrtf fuction twice but with this func we can avoid func call, maybe this is
+    not good name for this func
+
+    Parameters:
+      | *[in]*  **v**   vector
+
+    Returns:
+      square of norm / magnitude
+
+.. c:function:: float glm_vec2_norm(vec2 vec)
+
+    | euclidean norm (magnitude), also called L2 norm
+    | this will give magnitude of vector in euclidean space
+
+    Parameters:
+      | *[in]*  **vec**   vector
+
+.. c:function:: void glm_vec2_add(vec2 a, vec2 b, vec2 dest)
+
+    add a vector to b vector store result in dest
+
+    Parameters:
+      | *[in]*  **a**     vector1
+      | *[in]*  **b**     vector2
+      | *[out]* **dest**  destination vector
+
+.. c:function:: void glm_vec2_adds(vec2 a, float s, vec2 dest)
+
+    add scalar to v vector store result in dest (d = v + vec(s))
+
+    Parameters:
+      | *[in]*  **v**     vector
+      | *[in]*  **s**     scalar
+      | *[out]* **dest**  destination vector
+
+.. c:function:: void glm_vec2_sub(vec2 v1, vec2 v2, vec2 dest)
+
+    subtract b vector from a vector store result in dest (d = v1 - v2)
+
+    Parameters:
+      | *[in]*  **a**     vector1
+      | *[in]*  **b**     vector2
+      | *[out]* **dest**  destination vector
+
+.. c:function:: void glm_vec2_subs(vec2 v, float s, vec2 dest)
+
+    subtract scalar from v vector store result in dest (d = v - vec(s))
+
+    Parameters:
+      | *[in]*  **v**     vector
+      | *[in]*  **s**     scalar
+      | *[out]* **dest**  destination vector
+
+.. c:function:: void glm_vec2_mul(vec2 a, vec2 b, vec2 d)
+
+    multiply two vector (component-wise multiplication)
+
+    Parameters:
+      | *[in]*  **a**     vector
+      | *[in]*  **b**     scalar
+      | *[out]* **d**     result = (a[0] * b[0], a[1] * b[1], a[2] * b[2])
+
+.. c:function:: void glm_vec2_scale(vec2 v, float s, vec2 dest)
+
+     multiply/scale vec2 vector with scalar: result = v * s
+
+
+    Parameters:
+      | *[in]*  **v**     vector
+      | *[in]*  **s**     scalar
+      | *[out]* **dest**  destination vector
+
+.. c:function:: void glm_vec2_scale_as(vec2 v, float s, vec2 dest)
+
+    make vec2 vector scale as specified: result = unit(v) * s
+
+    Parameters:
+      | *[in]*  **v**     vector
+      | *[in]*  **s**     scalar
+      | *[out]* **dest**  destination vector
+
+.. c:function:: void glm_vec2_div(vec2 a, vec2 b, vec2 dest)
+
+    div vector with another component-wise division: d = a / b
+
+    Parameters:
+      | *[in]*  **a**     vector 1
+      | *[in]*  **b**     vector 2
+      | *[out]* **dest**  result = (a[0] / b[0], a[1] / b[1], a[2] / b[2])
+
+.. c:function:: void glm_vec2_divs(vec2 v, float s, vec2 dest)
+
+    div vector with scalar: d = v / s
+
+    Parameters:
+      | *[in]*  **v**     vector
+      | *[in]*  **s**     scalar
+      | *[out]* **dest**  result = (a[0] / s, a[1] / s, a[2] / s])
+
+.. c:function:: void glm_vec2_addadd(vec2 a, vec2 b, vec2 dest)
+
+    | add two vectors and add result to sum
+    | it applies += operator so dest must be initialized
+
+    Parameters:
+      | *[in]*  **a**     vector 1
+      | *[in]*  **b**     vector 2
+      | *[out]* **dest**  dest += (a + b)
+
+.. c:function:: void glm_vec2_subadd(vec2 a, vec2 b, vec2 dest)
+
+    | sub two vectors and add result to sum
+    | it applies += operator so dest must be initialized
+
+    Parameters:
+      | *[in]*  **a**     vector 1
+      | *[in]*  **b**     vector 2
+      | *[out]* **dest**  dest += (a - b)
+
+.. c:function:: void glm_vec2_muladd(vec2 a, vec2 b, vec2 dest)
+
+    | mul two vectors and add result to sum
+    | it applies += operator so dest must be initialized
+
+    Parameters:
+      | *[in]*  **a**     vector 1
+      | *[in]*  **b**     vector 2
+      | *[out]* **dest**  dest += (a * b)
+
+.. c:function:: void glm_vec2_muladds(vec2 a, float s, vec2 dest)
+
+    | mul vector with scalar and add result to sum
+    | it applies += operator so dest must be initialized
+
+    Parameters:
+      | *[in]*  **a**     vector
+      | *[in]*  **s**     scalar
+      | *[out]* **dest**  dest += (a * b)
+
+.. c:function:: void glm_vec2_maxadd(vec2 a, vec2 b, vec2 dest)
+
+    | add max of two vector to result/dest
+    | it applies += operator so dest must be initialized
+
+    Parameters:
+      | *[in]*  **a**     vector 1
+      | *[in]*  **b**     vector 2
+      | *[out]* **dest**  dest += (a * b)
+
+.. c:function:: void glm_vec2_minadd(vec2 a, vec2 b, vec2 dest)
+
+    | add min of two vector to result/dest
+    | it applies += operator so dest must be initialized
+
+    Parameters:
+      | *[in]*  **a**     vector 1
+      | *[in]*  **b**     vector 2
+      | *[out]* **dest**  dest += (a * b)
+
+.. c:function:: void glm_vec2_negate(vec2 v)
+
+    negate vector components
+
+    Parameters:
+      | *[in, out]*  **v**    vector
+
+.. c:function:: void glm_vec2_negate_to(vec2 v, vec2 dest)
+
+    negate vector components and store result in dest
+
+    Parameters:
+      | *[in]*  **v**       vector
+      | *[out]* **dest**    negated vector
+
+.. c:function:: void glm_vec2_normalize(vec2 v)
+
+    normalize vec2 and store result in same vec
+
+    Parameters:
+      | *[in, out]*  **v**    vector
+
+.. c:function:: void glm_vec2_normalize_to(vec2 vec, vec2 dest)
+
+     normalize vec2 to dest
+
+    Parameters:
+      | *[in]*   **vec**   source
+      | *[out]*  **dest**  destination
+
+.. c:function:: void glm_vec2_rotate(vec2 v, float angle, vec2 dest)
+
+     rotate vec2 around axis by angle using Rodrigues' rotation formula
+
+    Parameters:
+      | *[in]*  **v**      vector
+      | *[in]*  **axis**   axis vector
+      | *[out]* **dest**   destination
+
+.. c:function:: float glm_vec2_distance2(vec2 v1, vec2 v2)
+
+    squared distance between two vectors
+
+    Parameters:
+      | *[in]*  **mat**   vector1
+      | *[in]*  **row1**  vector2
+
+    Returns:
+      | squared distance (distance * distance)
+
+.. c:function:: float glm_vec2_distance(vec2 v1, vec2 v2)
+
+    distance between two vectors
+
+    Parameters:
+      | *[in]*  **mat**   vector1
+      | *[in]*  **row1**  vector2
+
+    Returns:
+      | distance
+
+.. c:function:: void glm_vec2_maxv(vec2 v1, vec2 v2, vec2 dest)
+
+    max values of vectors
+
+    Parameters:
+      | *[in]*  **v1**    vector1
+      | *[in]*  **v2**    vector2
+      | *[out]* **dest**  destination
+
+.. c:function:: void glm_vec2_minv(vec2 v1, vec2 v2, vec2 dest)
+
+    min values of vectors
+
+    Parameters:
+      | *[in]*  **v1**    vector1
+      | *[in]*  **v2**    vector2
+      | *[out]* **dest**  destination
+
+.. c:function:: void glm_vec2_clamp(vec2 v, float minVal, float maxVal)
+
+    constrain a value to lie between two further values
+
+    Parameters:
+      | *[in, out]*  **v**       vector
+      | *[in]*       **minVal**  minimum value
+      | *[in]*       **maxVal**  maximum value
+
+.. c:function:: void glm_vec2_lerp(vec2 from, vec2 to, float t, vec2 dest)
+
+    linear interpolation between two vector
+
+    | formula:  from + s * (to - from)
+
+    Parameters:
+      | *[in]*  **from**   from value
+      | *[in]*  **to**     to value
+      | *[in]*  **t**      interpolant (amount) clamped between 0 and 1
+      | *[out]* **dest**   destination

docs/source/vec3-ext.rst

@@ -0,0 +1,143 @@
+.. default-domain:: C
+
+vec3 extra
+==========
+
+Header: cglm/vec3-ext.h
+
+There are some functions are in called in extra header. These are called extra
+because they are not used like other functions in vec3.h in the future some of
+these functions ma be moved to vec3 header.
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Functions:
+
+1. :c:func:`glm_vec3_mulv`
+#. :c:func:`glm_vec3_broadcast`
+#. :c:func:`glm_vec3_eq`
+#. :c:func:`glm_vec3_eq_eps`
+#. :c:func:`glm_vec3_eq_all`
+#. :c:func:`glm_vec3_eqv`
+#. :c:func:`glm_vec3_eqv_eps`
+#. :c:func:`glm_vec3_max`
+#. :c:func:`glm_vec3_min`
+#. :c:func:`glm_vec3_isnan`
+#. :c:func:`glm_vec3_isinf`
+#. :c:func:`glm_vec3_isvalid`
+#. :c:func:`glm_vec3_sign`
+#. :c:func:`glm_vec3_sqrt`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: void  glm_vec3_mulv(vec3 a, vec3 b, vec3 d)
+
+    multiplies individual items
+
+    Parameters:
+      | *[in]*  **a**  vec1
+      | *[in]*  **b**  vec2
+      | *[out]* **d**  destination (v1[0] * v2[0], v1[1] * v2[1], v1[2] * v2[2])
+
+.. c:function:: void  glm_vec3_broadcast(float val, vec3 d)
+
+    fill a vector with specified value
+
+    Parameters:
+      | *[in]*  **val**   value
+      | *[out]* **dest**  destination
+
+.. c:function:: bool  glm_vec3_eq(vec3 v, float val)
+
+    check if vector is equal to value (without epsilon)
+
+    Parameters:
+      | *[in]*  **v**    vector
+      | *[in]*  **val**  value
+
+.. c:function:: bool  glm_vec3_eq_eps(vec3 v, float val)
+
+    check if vector is equal to value (with epsilon)
+
+    Parameters:
+      | *[in]*  **v**    vector
+      | *[in]*  **val**  value
+
+.. c:function:: bool  glm_vec3_eq_all(vec3 v)
+
+    check if vectors members are equal (without epsilon)
+
+    Parameters:
+      | *[in]*  **v**   vector
+
+.. c:function:: bool  glm_vec3_eqv(vec3 v1, vec3 v2)
+
+    check if vector is equal to another (without epsilon) vector
+
+    Parameters:
+      | *[in]*  **vec**   vector 1
+      | *[in]*  **vec**   vector 2
+
+.. c:function:: bool  glm_vec3_eqv_eps(vec3 v1, vec3 v2)
+
+    check if vector is equal to another (with epsilon)
+
+    Parameters:
+      | *[in]*  **v1**    vector1
+      | *[in]*  **v2**    vector2
+
+.. c:function:: float  glm_vec3_max(vec3 v)
+
+    max value of vector
+
+    Parameters:
+      | *[in]*  **v**    vector
+
+.. c:function:: float glm_vec3_min(vec3 v)
+
+     min value of vector
+
+    Parameters:
+      | *[in]*  **v**  vector
+
+.. c:function:: bool glm_vec3_isnan(vec3 v)
+
+    | check if one of items is NaN (not a number)
+    | you should only use this in DEBUG mode or very critical asserts
+
+    Parameters:
+      | *[in]*  **v**  vector
+
+.. c:function:: bool glm_vec3_isinf(vec3 v)
+
+    | check if one of items is INFINITY
+    | you should only use this in DEBUG mode or very critical asserts
+
+    Parameters:
+      | *[in]*  **v**  vector
+
+.. c:function:: bool glm_vec3_isvalid(vec3 v)
+
+    | check if all items are valid number
+    | you should only use this in DEBUG mode or very critical asserts
+
+    Parameters:
+      | *[in]*  **v**  vector
+
+.. c:function:: void glm_vec3_sign(vec3 v, vec3 dest)
+
+    get sign of 32 bit float as +1, -1, 0
+
+    Parameters:
+      | *[in]*   **v**     vector
+      | *[out]*  **dest**  sign vector (only keeps signs as -1, 0, -1)
+
+.. c:function:: void glm_vec3_sqrt(vec3 v, vec3 dest)
+
+    square root of each vector item
+
+    Parameters:
+      | *[in]*   **v**     vector
+      | *[out]*  **dest**  destination vector (sqrt(v))

docs/source/vec3.rst

@@ -0,0 +1,500 @@
+.. default-domain:: C
+
+vec3
+====
+
+Header: cglm/vec3.h
+
+ **Important:** *cglm* was used **glm_vec_** namespace for vec3 functions until
+ **v0.5.0**, since **v0.5.0** cglm uses **glm_vec3_** namespace for vec3.
+
+ Also `glm_vec3_flipsign` has been renamed to `glm_vec3_negate`
+
+We mostly use vectors in graphics math, to make writing code faster
+and easy to read, some *vec3* functions are aliased in global namespace.
+For instance :c:func:`glm_dot` is alias of :c:func:`glm_vec3_dot`,
+alias means inline wrapper here. There is no call verison of alias functions
+
+There are also functions for rotating *vec3* vector. **_m4**, **_m3** prefixes
+rotate *vec3* with matrix.
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Macros:
+
+1. glm_vec3_dup(v, dest)
+#. GLM_VEC3_ONE_INIT
+#. GLM_VEC3_ZERO_INIT
+#. GLM_VEC3_ONE
+#. GLM_VEC3_ZERO
+#. GLM_YUP
+#. GLM_ZUP
+#. GLM_XUP
+
+Functions:
+
+1. :c:func:`glm_vec3`
+#. :c:func:`glm_vec3_copy`
+#. :c:func:`glm_vec3_zero`
+#. :c:func:`glm_vec3_one`
+#. :c:func:`glm_vec3_dot`
+#. :c:func:`glm_vec3_norm2`
+#. :c:func:`glm_vec3_norm`
+#. :c:func:`glm_vec3_add`
+#. :c:func:`glm_vec3_adds`
+#. :c:func:`glm_vec3_sub`
+#. :c:func:`glm_vec3_subs`
+#. :c:func:`glm_vec3_mul`
+#. :c:func:`glm_vec3_scale`
+#. :c:func:`glm_vec3_scale_as`
+#. :c:func:`glm_vec3_div`
+#. :c:func:`glm_vec3_divs`
+#. :c:func:`glm_vec3_addadd`
+#. :c:func:`glm_vec3_subadd`
+#. :c:func:`glm_vec3_muladd`
+#. :c:func:`glm_vec3_muladds`
+#. :c:func:`glm_vec3_maxadd`
+#. :c:func:`glm_vec3_minadd`
+#. :c:func:`glm_vec3_flipsign`
+#. :c:func:`glm_vec3_flipsign_to`
+#. :c:func:`glm_vec3_inv`
+#. :c:func:`glm_vec3_inv_to`
+#. :c:func:`glm_vec3_negate`
+#. :c:func:`glm_vec3_negate_to`
+#. :c:func:`glm_vec3_normalize`
+#. :c:func:`glm_vec3_normalize_to`
+#. :c:func:`glm_vec3_cross`
+#. :c:func:`glm_vec3_crossn`
+#. :c:func:`glm_vec3_distance2`
+#. :c:func:`glm_vec3_distance`
+#. :c:func:`glm_vec3_angle`
+#. :c:func:`glm_vec3_rotate`
+#. :c:func:`glm_vec3_rotate_m4`
+#. :c:func:`glm_vec3_rotate_m3`
+#. :c:func:`glm_vec3_proj`
+#. :c:func:`glm_vec3_center`
+#. :c:func:`glm_vec3_maxv`
+#. :c:func:`glm_vec3_minv`
+#. :c:func:`glm_vec3_ortho`
+#. :c:func:`glm_vec3_clamp`
+#. :c:func:`glm_vec3_lerp`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: void  glm_vec3(vec4 v4, vec3 dest)
+
+    init vec3 using vec4
+
+    Parameters:
+      | *[in]*  **v4**    vector4
+      | *[out]* **dest**  destination
+
+.. c:function:: void  glm_vec3_copy(vec3 a, vec3 dest)
+
+    copy all members of [a] to [dest]
+
+    Parameters:
+      | *[in]*  **a**     source
+      | *[out]* **dest**  destination
+
+.. c:function:: void  glm_vec3_zero(vec3 v)
+
+    makes all members 0.0f (zero)
+
+    Parameters:
+      | *[in, out]*  **v**     vector
+
+.. c:function:: void  glm_vec3_one(vec3 v)
+
+    makes all members 1.0f (one)
+
+    Parameters:
+      | *[in, out]*  **v**     vector
+
+.. c:function:: float  glm_vec3_dot(vec3 a, vec3 b)
+
+    dot product of vec3
+
+    Parameters:
+      | *[in]*  **a**  vector1
+      | *[in]*  **b**  vector2
+
+    Returns:
+      dot product
+
+.. c:function:: void  glm_vec3_cross(vec3 a, vec3 b, vec3 d)
+
+    cross product of two vector (RH)
+
+    Parameters:
+      | *[in]*  **a**     vector 1
+      | *[in]*  **b**     vector 2
+      | *[out]* **dest**  destination
+
+.. c:function:: void  glm_vec3_crossn(vec3 a, vec3 b, vec3 dest)
+
+    cross product of two vector (RH) and normalize the result
+
+    Parameters:
+      | *[in]*  **a**     vector 1
+      | *[in]*  **b**     vector 2
+      | *[out]* **dest**  destination
+
+.. c:function:: float  glm_vec3_norm2(vec3 v)
+
+    norm * norm (magnitude) of vector
+
+    we can use this func instead of calling norm * norm, because it would call
+    sqrtf fuction twice but with this func we can avoid func call, maybe this is
+    not good name for this func
+
+    Parameters:
+      | *[in]*  **v**   vector
+
+    Returns:
+      square of norm / magnitude
+
+.. c:function:: float  glm_vec3_norm(vec3 vec)
+
+    | euclidean norm (magnitude), also called L2 norm
+    | this will give magnitude of vector in euclidean space
+
+    Parameters:
+      | *[in]*  **vec**   vector
+
+.. c:function:: void  glm_vec3_add(vec3 a, vec3 b, vec3 dest)
+
+    add a vector to b vector store result in dest
+
+    Parameters:
+      | *[in]*  **a**     vector1
+      | *[in]*  **b**     vector2
+      | *[out]* **dest**  destination vector
+
+.. c:function:: void  glm_vec3_adds(vec3 a, float s, vec3 dest)
+
+    add scalar to v vector store result in dest (d = v + vec(s))
+
+    Parameters:
+      | *[in]*  **v**     vector
+      | *[in]*  **s**     scalar
+      | *[out]* **dest**  destination vector
+
+.. c:function:: void  glm_vec3_sub(vec3 v1, vec3 v2, vec3 dest)
+
+    subtract b vector from a vector store result in dest (d = v1 - v2)
+
+    Parameters:
+      | *[in]*  **a**     vector1
+      | *[in]*  **b**     vector2
+      | *[out]* **dest**  destination vector
+
+.. c:function:: void  glm_vec3_subs(vec3 v, float s, vec3 dest)
+
+    subtract scalar from v vector store result in dest (d = v - vec(s))
+
+    Parameters:
+      | *[in]*  **v**     vector
+      | *[in]*  **s**     scalar
+      | *[out]* **dest**  destination vector
+
+.. c:function:: void  glm_vec3_mul(vec3 a, vec3 b, vec3 d)
+
+    multiply two vector (component-wise multiplication)
+
+    Parameters:
+      | *[in]*  **a**     vector
+      | *[in]*  **b**     scalar
+      | *[out]* **d**     result = (a[0] * b[0], a[1] * b[1], a[2] * b[2])
+
+.. c:function:: void glm_vec3_scale(vec3 v, float s, vec3 dest)
+
+     multiply/scale vec3 vector with scalar: result = v * s
+
+
+    Parameters:
+      | *[in]*  **v**     vector
+      | *[in]*  **s**     scalar
+      | *[out]* **dest**  destination vector
+
+.. c:function:: void  glm_vec3_scale_as(vec3 v, float s, vec3 dest)
+
+    make vec3 vector scale as specified: result = unit(v) * s
+
+    Parameters:
+      | *[in]*  **v**     vector
+      | *[in]*  **s**     scalar
+      | *[out]* **dest**  destination vector
+
+.. c:function:: void  glm_vec3_div(vec3 a, vec3 b, vec3 dest)
+
+    div vector with another component-wise division: d = a / b
+
+    Parameters:
+      | *[in]*  **a**     vector 1
+      | *[in]*  **b**     vector 2
+      | *[out]* **dest**  result = (a[0] / b[0], a[1] / b[1], a[2] / b[2])
+
+.. c:function:: void  glm_vec3_divs(vec3 v, float s, vec3 dest)
+
+    div vector with scalar: d = v / s
+
+    Parameters:
+      | *[in]*  **v**     vector
+      | *[in]*  **s**     scalar
+      | *[out]* **dest**  result = (a[0] / s, a[1] / s, a[2] / s])
+
+.. c:function:: void  glm_vec3_addadd(vec3 a, vec3 b, vec3 dest)
+
+    | add two vectors and add result to sum
+    | it applies += operator so dest must be initialized
+
+    Parameters:
+      | *[in]*  **a**     vector 1
+      | *[in]*  **b**     vector 2
+      | *[out]* **dest**  dest += (a + b)
+
+.. c:function:: void  glm_vec3_subadd(vec3 a, vec3 b, vec3 dest)
+
+    | sub two vectors and add result to sum
+    | it applies += operator so dest must be initialized
+
+    Parameters:
+      | *[in]*  **a**     vector 1
+      | *[in]*  **b**     vector 2
+      | *[out]* **dest**  dest += (a - b)
+
+.. c:function:: void  glm_vec3_muladd(vec3 a, vec3 b, vec3 dest)
+
+    | mul two vectors and add result to sum
+    | it applies += operator so dest must be initialized
+
+    Parameters:
+      | *[in]*  **a**     vector 1
+      | *[in]*  **b**     vector 2
+      | *[out]* **dest**  dest += (a * b)
+
+.. c:function:: void  glm_vec3_muladds(vec3 a, float s, vec3 dest)
+
+    | mul vector with scalar and add result to sum
+    | it applies += operator so dest must be initialized
+
+    Parameters:
+      | *[in]*  **a**     vector
+      | *[in]*  **s**     scalar
+      | *[out]* **dest**  dest += (a * b)
+
+.. c:function:: void  glm_vec3_maxadd(vec3 a, vec3 b, vec3 dest)
+
+    | add max of two vector to result/dest
+    | it applies += operator so dest must be initialized
+
+    Parameters:
+      | *[in]*  **a**     vector 1
+      | *[in]*  **b**     vector 2
+      | *[out]* **dest**  dest += (a * b)
+
+.. c:function:: void  glm_vec3_minadd(vec3 a, vec3 b, vec3 dest)
+
+    | add min of two vector to result/dest
+    | it applies += operator so dest must be initialized
+
+    Parameters:
+      | *[in]*  **a**     vector 1
+      | *[in]*  **b**     vector 2
+      | *[out]* **dest**  dest += (a * b)
+
+.. c:function:: void  glm_vec3_flipsign(vec3 v)
+
+    **DEPRACATED!**
+
+    use :c:func:`glm_vec3_negate`
+
+    Parameters:
+      | *[in, out]*  **v**    vector
+
+.. c:function:: void  glm_vec3_flipsign_to(vec3 v, vec3 dest)
+
+    **DEPRACATED!**
+
+    use :c:func:`glm_vec3_negate_to`
+
+    Parameters:
+      | *[in]*  **v**       vector
+      | *[out]* **dest**    negated vector
+
+.. c:function:: void  glm_vec3_inv(vec3 v)
+
+    **DEPRACATED!**
+
+    use :c:func:`glm_vec3_negate`
+
+    Parameters:
+      | *[in, out]*  **v**    vector
+
+.. c:function:: void  glm_vec3_inv_to(vec3 v, vec3 dest)
+
+    **DEPRACATED!**
+
+    use :c:func:`glm_vec3_negate_to`
+
+    Parameters:
+      | *[in]*  **v**     source
+      | *[out]* **dest**  destination
+
+.. c:function:: void  glm_vec3_negate(vec3 v)
+
+    negate vector components
+
+    Parameters:
+      | *[in, out]*  **v**    vector
+
+.. c:function:: void  glm_vec3_negate_to(vec3 v, vec3 dest)
+
+    negate vector components and store result in dest
+
+    Parameters:
+      | *[in]*  **v**       vector
+      | *[out]* **dest**    negated vector
+
+.. c:function:: void  glm_vec3_normalize(vec3 v)
+
+    normalize vec3 and store result in same vec
+
+    Parameters:
+      | *[in, out]*  **v**    vector
+
+.. c:function:: void  glm_vec3_normalize_to(vec3 vec, vec3 dest)
+
+     normalize vec3 to dest
+
+    Parameters:
+      | *[in]*   **vec**   source
+      | *[out]*  **dest**  destination
+
+.. c:function:: float  glm_vec3_angle(vec3 v1, vec3 v2)
+
+    angle betwen two vector
+
+    Parameters:
+      | *[in]*  **v1**   vector1
+      | *[in]*  **v2**   vector2
+
+    Return:
+      | angle as radians
+
+.. c:function:: void  glm_vec3_rotate(vec3 v, float angle, vec3 axis)
+
+     rotate vec3 around axis by angle using Rodrigues' rotation formula
+
+    Parameters:
+      | *[in, out]*  **v**      vector
+      | *[in]*       **axis**   axis vector (will be normalized)
+      | *[out]*      **angle**  angle (radians)
+
+.. c:function:: void  glm_vec3_rotate_m4(mat4 m, vec3 v, vec3 dest)
+
+    apply rotation matrix to vector
+
+    Parameters:
+      | *[in]*  **m**     affine matrix or rot matrix
+      | *[in]*  **v**     vector
+      | *[out]* **dest**  rotated vector
+
+.. c:function:: void  glm_vec3_rotate_m3(mat3 m, vec3 v, vec3 dest)
+
+    apply rotation matrix to vector
+
+    Parameters:
+      | *[in]*  **m**     affine matrix or rot matrix
+      | *[in]*  **v**     vector
+      | *[out]* **dest**  rotated vector
+
+.. c:function:: void  glm_vec3_proj(vec3 a, vec3 b, vec3 dest)
+
+    project a vector onto b vector
+
+    Parameters:
+      | *[in]*  **a**     vector1
+      | *[in]*  **b**     vector2
+      | *[out]* **dest**  projected vector
+
+.. c:function:: void  glm_vec3_center(vec3 v1, vec3 v2, vec3 dest)
+
+    find center point of two vector
+
+    Parameters:
+      | *[in]*  **v1**    vector1
+      | *[in]*  **v2**    vector2
+      | *[out]* **dest**  center point
+
+.. c:function:: float  glm_vec3_distance2(vec3 v1, vec3 v2)
+
+    squared distance between two vectors
+
+    Parameters:
+      | *[in]*  **mat**   vector1
+      | *[in]*  **row1**  vector2
+
+    Returns:
+      | squared distance (distance * distance)
+
+.. c:function:: float  glm_vec3_distance(vec3 v1, vec3 v2)
+
+    distance between two vectors
+
+    Parameters:
+      | *[in]*  **mat**   vector1
+      | *[in]*  **row1**  vector2
+
+    Returns:
+      | distance
+
+.. c:function:: void  glm_vec3_maxv(vec3 v1, vec3 v2, vec3 dest)
+
+    max values of vectors
+
+    Parameters:
+      | *[in]*  **v1**    vector1
+      | *[in]*  **v2**    vector2
+      | *[out]* **dest**  destination
+
+.. c:function:: void  glm_vec3_minv(vec3 v1, vec3 v2, vec3 dest)
+
+    min values of vectors
+
+    Parameters:
+      | *[in]*  **v1**    vector1
+      | *[in]*  **v2**    vector2
+      | *[out]* **dest**  destination
+
+.. c:function:: void  glm_vec3_ortho(vec3 v, vec3 dest)
+
+    possible orthogonal/perpendicular vector
+
+    Parameters:
+      | *[in]*  **mat**   vector
+      | *[out]* **dest**  orthogonal/perpendicular vector
+
+.. c:function:: void  glm_vec3_clamp(vec3 v, float minVal, float maxVal)
+
+    constrain a value to lie between two further values
+
+    Parameters:
+      | *[in, out]*  **v**       vector
+      | *[in]*       **minVal**  minimum value
+      | *[in]*       **maxVal**  maximum value
+
+.. c:function:: void  glm_vec3_lerp(vec3 from, vec3 to, float t, vec3 dest)
+
+    linear interpolation between two vector
+
+    | formula:  from + s * (to - from)
+
+    Parameters:
+      | *[in]*  **from**   from value
+      | *[in]*  **to**     to value
+      | *[in]*  **t**      interpolant (amount) clamped between 0 and 1
+      | *[out]* **dest**   destination

docs/source/vec4-ext.rst

@@ -0,0 +1,138 @@
+.. default-domain:: C
+
+vec4 extra
+==========
+
+Header: cglm/vec4-ext.h
+
+There are some functions are in called in extra header. These are called extra
+because they are not used like other functions in vec4.h in the future some of
+these functions ma be moved to vec4 header.
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Functions:
+
+1. :c:func:`glm_vec4_mulv`
+#. :c:func:`glm_vec4_broadcast`
+#. :c:func:`glm_vec4_eq`
+#. :c:func:`glm_vec4_eq_eps`
+#. :c:func:`glm_vec4_eq_all`
+#. :c:func:`glm_vec4_eqv`
+#. :c:func:`glm_vec4_eqv_eps`
+#. :c:func:`glm_vec4_max`
+#. :c:func:`glm_vec4_min`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: void  glm_vec4_mulv(vec4 a, vec4 b, vec4 d)
+
+    multiplies individual items
+
+    Parameters:
+      | *[in]*  **a**  vec1
+      | *[in]*  **b**  vec2
+      | *[out]* **d**  destination
+
+.. c:function:: void  glm_vec4_broadcast(float val, vec4 d)
+
+    fill a vector with specified value
+
+    Parameters:
+      | *[in]*  **val**   value
+      | *[out]* **dest**  destination
+
+.. c:function:: bool  glm_vec4_eq(vec4 v, float val)
+
+    check if vector is equal to value (without epsilon)
+
+    Parameters:
+      | *[in]*  **v**    vector
+      | *[in]*  **val**  value
+
+.. c:function:: bool  glm_vec4_eq_eps(vec4 v, float val)
+
+    check if vector is equal to value (with epsilon)
+
+    Parameters:
+      | *[in]*  **v**    vector
+      | *[in]*  **val**  value
+
+.. c:function:: bool  glm_vec4_eq_all(vec4 v)
+
+    check if vectors members are equal (without epsilon)
+
+    Parameters:
+      | *[in]*  **v**   vector
+
+.. c:function:: bool  glm_vec4_eqv(vec4 v1, vec4 v2)
+
+    check if vector is equal to another (without epsilon) vector
+
+    Parameters:
+      | *[in]*  **vec**   vector 1
+      | *[in]*  **vec**   vector 2
+
+.. c:function:: bool  glm_vec4_eqv_eps(vec4 v1, vec4 v2)
+
+    check if vector is equal to another (with epsilon)
+
+    Parameters:
+      | *[in]*  **v1**    vector1
+      | *[in]*  **v2**    vector2
+
+.. c:function:: float  glm_vec4_max(vec4 v)
+
+    max value of vector
+
+    Parameters:
+      | *[in]*  **v**    vector
+
+.. c:function:: float glm_vec4_min(vec4 v)
+
+     min value of vector
+
+    Parameters:
+      | *[in]*  **v**  vector
+
+.. c:function:: bool glm_vec4_isnan(vec4 v)
+
+    | check if one of items is NaN (not a number)
+    | you should only use this in DEBUG mode or very critical asserts
+
+    Parameters:
+      | *[in]*  **v**  vector
+
+.. c:function:: bool glm_vec4_isinf(vec4 v)
+
+    | check if one of items is INFINITY
+    | you should only use this in DEBUG mode or very critical asserts
+
+    Parameters:
+      | *[in]*  **v**  vector
+
+.. c:function:: bool glm_vec4_isvalid(vec4 v)
+
+    | check if all items are valid number
+    | you should only use this in DEBUG mode or very critical asserts
+
+    Parameters:
+      | *[in]*  **v**  vector
+
+.. c:function:: void glm_vec4_sign(vec4 v, vec4 dest)
+
+    get sign of 32 bit float as +1, -1, 0
+
+    Parameters:
+      | *[in]*   **v**     vector
+      | *[out]*  **dest**  sign vector (only keeps signs as -1, 0, -1)
+
+.. c:function:: void glm_vec4_sqrt(vec4 v, vec4 dest)
+
+    square root of each vector item
+
+    Parameters:
+      | *[in]*   **v**     vector
+      | *[out]*  **dest**  destination vector (sqrt(v))

docs/source/vec4.rst

@@ -0,0 +1,408 @@
+.. default-domain:: C
+
+vec4
+====
+
+Header: cglm/vec4.h
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Macros:
+
+1. glm_vec4_dup3(v, dest)
+#. glm_vec4_dup(v, dest)
+#. GLM_VEC4_ONE_INIT
+#. GLM_VEC4_BLACK_INIT
+#. GLM_VEC4_ZERO_INIT
+#. GLM_VEC4_ONE
+#. GLM_VEC4_BLACK
+#. GLM_VEC4_ZERO
+
+Functions:
+
+1. :c:func:`glm_vec4`
+#. :c:func:`glm_vec4_copy3`
+#. :c:func:`glm_vec4_copy`
+#. :c:func:`glm_vec4_ucopy`
+#. :c:func:`glm_vec4_zero`
+#. :c:func:`glm_vec4_one`
+#. :c:func:`glm_vec4_dot`
+#. :c:func:`glm_vec4_norm2`
+#. :c:func:`glm_vec4_norm`
+#. :c:func:`glm_vec4_add`
+#. :c:func:`glm_vec4_adds`
+#. :c:func:`glm_vec4_sub`
+#. :c:func:`glm_vec4_subs`
+#. :c:func:`glm_vec4_mul`
+#. :c:func:`glm_vec4_scale`
+#. :c:func:`glm_vec4_scale_as`
+#. :c:func:`glm_vec4_div`
+#. :c:func:`glm_vec4_divs`
+#. :c:func:`glm_vec4_addadd`
+#. :c:func:`glm_vec4_subadd`
+#. :c:func:`glm_vec4_muladd`
+#. :c:func:`glm_vec4_muladds`
+#. :c:func:`glm_vec4_maxadd`
+#. :c:func:`glm_vec4_minadd`
+#. :c:func:`glm_vec4_flipsign`
+#. :c:func:`glm_vec4_flipsign_to`
+#. :c:func:`glm_vec4_inv`
+#. :c:func:`glm_vec4_inv_to`
+#. :c:func:`glm_vec4_negate`
+#. :c:func:`glm_vec4_negate_to`
+#. :c:func:`glm_vec4_normalize`
+#. :c:func:`glm_vec4_normalize_to`
+#. :c:func:`glm_vec4_distance`
+#. :c:func:`glm_vec4_maxv`
+#. :c:func:`glm_vec4_minv`
+#. :c:func:`glm_vec4_clamp`
+#. :c:func:`glm_vec4_lerp`
+#. :c:func:`glm_vec4_cubic`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: void  glm_vec4(vec3 v3, float last, vec4 dest)
+
+    init vec4 using vec3, since you are initializing vec4 with vec3
+    you need to set last item. cglm could set it zero but making it parameter
+    gives more control
+
+    Parameters:
+      | *[in]*  **v3**    vector4
+      | *[in]*  **last**  last item of vec4
+      | *[out]* **dest**  destination
+
+.. c:function:: void glm_vec4_copy3(vec4 a, vec3 dest)
+
+    copy first 3 members of [a] to [dest]
+
+    Parameters:
+      | *[in]*  **a**     source
+      | *[out]* **dest**  destination
+
+.. c:function:: void  glm_vec4_copy(vec4 v, vec4 dest)
+
+    copy all members of [a] to [dest]
+
+    Parameters:
+      | *[in]*  **v**     source
+      | *[in]*  **dest**  destination
+
+.. c:function:: void  glm_vec4_ucopy(vec4 v, vec4 dest)
+
+    copy all members of [a] to [dest]
+
+    | alignment is not required
+
+    Parameters:
+      | *[in]*  **v**     source
+      | *[in]*  **dest**  destination
+
+.. c:function:: void  glm_vec4_zero(vec4 v)
+
+    makes all members zero
+
+    Parameters:
+      | *[in, out]*  **v**     vector
+
+.. c:function:: float  glm_vec4_dot(vec4 a, vec4 b)
+
+    dot product of vec4
+
+    Parameters:
+      | *[in]*  **a**  vector1
+      | *[in]*  **b**  vector2
+
+    Returns:
+      dot product
+
+.. c:function:: float  glm_vec4_norm2(vec4 v)
+
+    norm * norm (magnitude) of vector
+
+    we can use this func instead of calling norm * norm, because it would call
+    sqrtf fuction twice but with this func we can avoid func call, maybe this is
+    not good name for this func
+
+    Parameters:
+      | *[in]*  **v**   vector
+
+    Returns:
+      square of norm / magnitude
+
+.. c:function:: float  glm_vec4_norm(vec4 vec)
+
+    | euclidean norm (magnitude), also called L2 norm
+    | this will give magnitude of vector in euclidean space
+
+    Parameters:
+      | *[in]*  **vec**   vector
+
+.. c:function:: void  glm_vec4_add(vec4 a, vec4 b, vec4 dest)
+
+    add a vector to b vector store result in dest
+
+    Parameters:
+      | *[in]*  **a**     vector1
+      | *[in]*  **b**     vector2
+      | *[out]* **dest**  destination vector
+
+.. c:function:: void  glm_vec4_adds(vec4 v, float s, vec4 dest)
+
+    add scalar to v vector store result in dest (d = v + vec(s))
+
+    Parameters:
+      | *[in]*  **v**     vector
+      | *[in]*  **s**     scalar
+      | *[out]* **dest**  destination vector
+
+.. c:function:: void  glm_vec4_sub(vec4 a, vec4 b, vec4 dest)
+
+    subtract b vector from a vector store result in dest (d = v1 - v2)
+
+    Parameters:
+      | *[in]*  **a**     vector1
+      | *[in]*  **b**     vector2
+      | *[out]* **dest**  destination vector
+
+.. c:function:: void  glm_vec4_subs(vec4 v, float s, vec4 dest)
+
+    subtract scalar from v vector store result in dest (d = v - vec(s))
+
+    Parameters:
+      | *[in]*  **v**     vector
+      | *[in]*  **s**     scalar
+      | *[out]* **dest**  destination vector
+
+.. c:function:: void  glm_vec4_mul(vec4 a, vec4 b, vec4 d)
+
+    multiply two vector (component-wise multiplication)
+
+    Parameters:
+      | *[in]*  **a**     vector1
+      | *[in]*  **b**     vector2
+      | *[out]* **dest**  result = (a[0] * b[0], a[1] * b[1], a[2] * b[2], a[3] * b[3])
+
+.. c:function:: void glm_vec4_scale(vec4 v, float s, vec4 dest)
+
+     multiply/scale vec4 vector with scalar: result = v * s
+
+    Parameters:
+      | *[in]*  **v**     vector
+      | *[in]*  **s**     scalar
+      | *[out]* **dest**  destination vector
+
+.. c:function:: void  glm_vec4_scale_as(vec4 v, float s, vec4 dest)
+
+    make vec4 vector scale as specified: result = unit(v) * s
+
+    Parameters:
+      | *[in]*  **v**     vector
+      | *[in]*  **s**     scalar
+      | *[out]* **dest**  destination vector
+
+.. c:function:: void  glm_vec4_div(vec4 a, vec4 b, vec4 dest)
+
+    div vector with another component-wise division: d = v1 / v2
+
+    Parameters:
+      | *[in]*  **a**     vector1
+      | *[in]*  **b**     vector2
+      | *[out]* **dest**  result = (a[0] / b[0], a[1] / b[1], a[2] / b[2], a[3] / b[3])
+
+.. c:function:: void  glm_vec4_divs(vec4 v, float s, vec4 dest)
+
+    div vector with scalar: d = v / s
+
+    Parameters:
+      | *[in]*  **v**     vector
+      | *[in]*  **s**     scalar
+      | *[out]* **dest**  result = (a[0] / s, a[1] / s, a[2] / s, a[3] / s)
+
+.. c:function:: void  glm_vec4_addadd(vec4 a, vec4 b, vec4 dest)
+
+    | add two vectors and add result to sum
+    | it applies += operator so dest must be initialized
+
+    Parameters:
+      | *[in]*  **a**     vector 1
+      | *[in]*  **b**     vector 2
+      | *[out]* **dest**  dest += (a + b)
+
+.. c:function:: void  glm_vec4_subadd(vec4 a, vec4 b, vec4 dest)
+
+    | sub two vectors and add result to sum
+    | it applies += operator so dest must be initialized
+
+    Parameters:
+      | *[in]*  **a**     vector 1
+      | *[in]*  **b**     vector 2
+      | *[out]* **dest**  dest += (a - b)
+
+.. c:function:: void  glm_vec4_muladd(vec4 a, vec4 b, vec4 dest)
+
+    | mul two vectors and add result to sum
+    | it applies += operator so dest must be initialized
+
+    Parameters:
+      | *[in]*  **a**     vector 1
+      | *[in]*  **b**     vector 2
+      | *[out]* **dest**  dest += (a * b)
+
+.. c:function:: void  glm_vec4_muladds(vec4 a, float s, vec4 dest)
+
+    | mul vector with scalar and add result to sum
+    | it applies += operator so dest must be initialized
+
+    Parameters:
+      | *[in]*  **a**     vector
+      | *[in]*  **s**     scalar
+      | *[out]* **dest**  dest += (a * b)
+
+.. c:function:: void  glm_vec4_maxadd(vec4 a, vec4 b, vec4 dest)
+
+    | add max of two vector to result/dest
+    | it applies += operator so dest must be initialized
+
+    Parameters:
+      | *[in]*  **a**     vector 1
+      | *[in]*  **b**     vector 2
+      | *[out]* **dest**  dest += (a * b)
+
+.. c:function:: void  glm_vec4_minadd(vec4 a, vec4 b, vec4 dest)
+
+    | add min of two vector to result/dest
+    | it applies += operator so dest must be initialized
+
+    Parameters:
+      | *[in]*  **a**     vector 1
+      | *[in]*  **b**     vector 2
+      | *[out]* **dest**  dest += (a * b)
+
+.. c:function:: void  glm_vec4_flipsign(vec4 v)
+
+    **DEPRACATED!**
+
+    use :c:func:`glm_vec4_negate`
+
+    Parameters:
+    | *[in, out]*  **v**    vector
+
+.. c:function:: void  glm_vec4_flipsign_to(vec4 v, vec4 dest)
+
+    **DEPRACATED!**
+
+    use :c:func:`glm_vec4_negate_to`
+
+    Parameters:
+      | *[in]*  **v**       vector
+      | *[out]* **dest**    negated vector
+
+.. c:function:: void  glm_vec4_inv(vec4 v)
+
+    **DEPRACATED!**
+
+    use :c:func:`glm_vec4_negate`
+
+    Parameters:
+      | *[in, out]*  **v**    vector
+
+.. c:function:: void  glm_vec4_inv_to(vec4 v, vec4 dest)
+
+    **DEPRACATED!**
+
+    use :c:func:`glm_vec4_negate_to`
+
+    Parameters:
+      | *[in]*  **v**     source
+      | *[out]* **dest**  destination
+
+.. c:function:: void  glm_vec4_negate(vec4 v)
+
+    negate vector components
+
+    Parameters:
+    | *[in, out]*  **v**    vector
+
+.. c:function:: void  glm_vec4_negate_to(vec4 v, vec4 dest)
+
+    negate vector components and store result in dest
+
+    Parameters:
+      | *[in]*  **v**       vector
+      | *[out]* **dest**    negated vector
+
+.. c:function:: void  glm_vec4_normalize(vec4 v)
+
+    normalize vec4 and store result in same vec
+
+    Parameters:
+      | *[in, out]*  **v**    vector
+
+.. c:function:: void  glm_vec4_normalize_to(vec4 vec, vec4 dest)
+
+    normalize vec4 to dest
+
+    Parameters:
+      | *[in]*   **vec**   source
+      | *[out]*  **dest**  destination
+
+.. c:function:: float  glm_vec4_distance(vec4 v1, vec4 v2)
+
+    distance between two vectors
+
+    Parameters:
+      | *[in]*  **mat**   vector1
+      | *[in]*  **row1**  vector2
+
+    Returns:
+      | distance
+
+.. c:function:: void  glm_vec4_maxv(vec4 v1, vec4 v2, vec4 dest)
+
+    max values of vectors
+
+    Parameters:
+      | *[in]*  **v1**    vector1
+      | *[in]*  **v2**    vector2
+      | *[out]* **dest**  destination
+
+.. c:function:: void  glm_vec4_minv(vec4 v1, vec4 v2, vec4 dest)
+
+    min values of vectors
+
+    Parameters:
+      | *[in]*  **v1**    vector1
+      | *[in]*  **v2**    vector2
+      | *[out]* **dest**  destination
+
+.. c:function:: void  glm_vec4_clamp(vec4 v, float minVal, float maxVal)
+
+    constrain a value to lie between two further values
+
+    Parameters:
+      | *[in, out]*  **v**       vector
+      | *[in]*       **minVal**  minimum value
+      | *[in]*       **maxVal**  maximum value
+
+.. c:function:: void  glm_vec4_lerp(vec4 from, vec4 to, float t, vec4 dest)
+
+    linear interpolation between two vector
+
+    | formula:  from + s * (to - from)
+
+    Parameters:
+      | *[in]*  **from**   from value
+      | *[in]*  **to**     to value
+      | *[in]*  **t**      interpolant (amount) clamped between 0 and 1
+      | *[out]* **dest**   destination
+
+.. c:function:: void  glm_vec4_cubic(float s, vec4 dest)
+
+    helper to fill vec4 as [S^3, S^2, S, 1]
+
+    Parameters:
+      | *[in]*  **s**      parameter
+      | *[out]* **dest**   destination

include/cglm/affine-mat.h

@@ -16,6 +16,7 @@
 
 #include "common.h"
 #include "mat4.h"
+#include "mat3.h"
 
 #ifdef CGLM_SSE_FP
 #  include "simd/sse2/affine.h"
@@ -25,8 +26,22 @@
 #  include "simd/avx/affine.h"
 #endif
 
-#include <assert.h>
-
+/*!
+ * @brief this is similar to glm_mat4_mul but specialized to affine transform
+ *
+ * Matrix format should be:
+ *   R  R  R  X
+ *   R  R  R  Y
+ *   R  R  R  Z
+ *   0  0  0  W
+ *
+ * this reduces some multiplications. It should be faster than mat4_mul.
+ * if you are not sure about matrix format then DON'T use this! use mat4_mul
+ *
+ * @param[in]   m1    affine matrix 1
+ * @param[in]   m2    affine matrix 2
+ * @param[out]  dest  result matrix
+ */
 CGLM_INLINE
 void
 glm_mul(mat4 m1, mat4 m2, mat4 dest) {
@@ -67,6 +82,59 @@ glm_mul(mat4 m1, mat4 m2, mat4 dest) {
 #endif
 }
 
+/*!
+ * @brief this is similar to glm_mat4_mul but specialized to affine transform
+ *
+ * Right Matrix format should be:
+ *   R  R  R  0
+ *   R  R  R  0
+ *   R  R  R  0
+ *   0  0  0  1
+ *
+ * this reduces some multiplications. It should be faster than mat4_mul.
+ * if you are not sure about matrix format then DON'T use this! use mat4_mul
+ *
+ * @param[in]   m1    affine matrix 1
+ * @param[in]   m2    affine matrix 2
+ * @param[out]  dest  result matrix
+ */
+CGLM_INLINE
+void
+glm_mul_rot(mat4 m1, mat4 m2, mat4 dest) {
+#if defined( __SSE__ ) || defined( __SSE2__ )
+  glm_mul_rot_sse2(m1, m2, dest);
+#else
+  float a00 = m1[0][0], a01 = m1[0][1], a02 = m1[0][2], a03 = m1[0][3],
+        a10 = m1[1][0], a11 = m1[1][1], a12 = m1[1][2], a13 = m1[1][3],
+        a20 = m1[2][0], a21 = m1[2][1], a22 = m1[2][2], a23 = m1[2][3],
+        a30 = m1[3][0], a31 = m1[3][1], a32 = m1[3][2], a33 = m1[3][3],
+
+        b00 = m2[0][0], b01 = m2[0][1], b02 = m2[0][2],
+        b10 = m2[1][0], b11 = m2[1][1], b12 = m2[1][2],
+        b20 = m2[2][0], b21 = m2[2][1], b22 = m2[2][2];
+
+  dest[0][0] = a00 * b00 + a10 * b01 + a20 * b02;
+  dest[0][1] = a01 * b00 + a11 * b01 + a21 * b02;
+  dest[0][2] = a02 * b00 + a12 * b01 + a22 * b02;
+  dest[0][3] = a03 * b00 + a13 * b01 + a23 * b02;
+
+  dest[1][0] = a00 * b10 + a10 * b11 + a20 * b12;
+  dest[1][1] = a01 * b10 + a11 * b11 + a21 * b12;
+  dest[1][2] = a02 * b10 + a12 * b11 + a22 * b12;
+  dest[1][3] = a03 * b10 + a13 * b11 + a23 * b12;
+
+  dest[2][0] = a00 * b20 + a10 * b21 + a20 * b22;
+  dest[2][1] = a01 * b20 + a11 * b21 + a21 * b22;
+  dest[2][2] = a02 * b20 + a12 * b21 + a22 * b22;
+  dest[2][3] = a03 * b20 + a13 * b21 + a23 * b22;
+
+  dest[3][0] = a30;
+  dest[3][1] = a31;
+  dest[3][2] = a32;
+  dest[3][3] = a33;
+#endif
+}
+
 /*!
  * @brief inverse orthonormal rotation + translation matrix (ridig-body)
  *
@@ -83,8 +151,8 @@ glm_inv_tr(mat4 mat) {
 #if defined( __SSE__ ) || defined( __SSE2__ )
   glm_inv_tr_sse2(mat);
 #else
-  CGLM_ALIGN(16) mat3 r;
-  CGLM_ALIGN(16) vec3 t;
+  CGLM_ALIGN_MAT mat3 r;
+  CGLM_ALIGN(8)  vec3 t;
 
   /* rotate */
   glm_mat4_pick3t(mat, r);
@@ -92,8 +160,8 @@ glm_inv_tr(mat4 mat) {
 
   /* translate */
   glm_mat3_mulv(r, mat[3], t);
-  glm_vec_flipsign(t);
-  glm_vec_copy(t, mat[3]);
+  glm_vec3_negate(t);
+  glm_vec3_copy(t, mat[3]);
 #endif
 }
 

include/cglm/affine.h

@@ -16,14 +16,14 @@
    CGLM_INLINE void glm_scale_to(mat4 m, vec3 v, mat4 dest);
    CGLM_INLINE void glm_scale_make(mat4 m, vec3 v);
    CGLM_INLINE void glm_scale(mat4 m, vec3 v);
-   CGLM_INLINE void glm_scale1(mat4 m, float s);
-   CGLM_INLINE void glm_rotate_x(mat4 m, float rad, mat4 dest);
-   CGLM_INLINE void glm_rotate_y(mat4 m, float rad, mat4 dest);
-   CGLM_INLINE void glm_rotate_z(mat4 m, float rad, mat4 dest);
-   CGLM_INLINE void glm_rotate_ndc_make(mat4 m, float angle, vec3 axis_ndc);
+   CGLM_INLINE void glm_scale_uni(mat4 m, float s);
+   CGLM_INLINE void glm_rotate_x(mat4 m, float angle, mat4 dest);
+   CGLM_INLINE void glm_rotate_y(mat4 m, float angle, mat4 dest);
+   CGLM_INLINE void glm_rotate_z(mat4 m, float angle, mat4 dest);
    CGLM_INLINE void glm_rotate_make(mat4 m, float angle, vec3 axis);
-   CGLM_INLINE void glm_rotate_ndc(mat4 m, float angle, vec3 axis_ndc);
    CGLM_INLINE void glm_rotate(mat4 m, float angle, vec3 axis);
+   CGLM_INLINE void glm_rotate_at(mat4 m, vec3 pivot, float angle, vec3 axis);
+   CGLM_INLINE void glm_rotate_atm(mat4 m, vec3 pivot, float angle, vec3 axis);
    CGLM_INLINE void glm_decompose_scalev(mat4 m, vec3 s);
    CGLM_INLINE bool glm_uniscaled(mat4 m);
    CGLM_INLINE void glm_decompose_rs(mat4 m, mat4 r, vec3 s);
@@ -34,56 +34,35 @@
 #define cglm_affine_h
 
 #include "common.h"
-#include "vec4.h"
-#include "affine-mat.h"
 #include "util.h"
+#include "vec3.h"
+#include "vec4.h"
+#include "mat4.h"
+#include "affine-mat.h"
 
 CGLM_INLINE
 void
-glm_translate_to(mat4 m, vec3 v, mat4 dest) {
-  mat4 t = GLM_MAT4_IDENTITY_INIT;
-
-#if defined( __SSE__ ) || defined( __SSE2__ )
-  _mm_store_ps(dest[3],
-               _mm_add_ps(_mm_add_ps(_mm_mul_ps(_mm_load_ps(t[0]),
-                                                _mm_set1_ps(v[0])),
-                                     _mm_mul_ps(_mm_load_ps(t[1]),
-                                                _mm_set1_ps(v[1]))),
-                          _mm_add_ps(_mm_mul_ps(_mm_load_ps(t[2]),
-                                                _mm_set1_ps(v[2])),
-                                     _mm_load_ps(t[3]))))
-  ;
-
-  _mm_store_ps(dest[0], _mm_load_ps(m[0]));
-  _mm_store_ps(dest[1], _mm_load_ps(m[1]));
-  _mm_store_ps(dest[2], _mm_load_ps(m[2]));
-#else
-  vec4 v1, v2, v3;
-
-  glm_vec4_scale(t[0], v[0], v1);
-  glm_vec4_scale(t[1], v[1], v2);
-  glm_vec4_scale(t[2], v[2], v3);
-
-  glm_vec4_add(v1, t[3], t[3]);
-  glm_vec4_add(v2, t[3], t[3]);
-  glm_vec4_add(v3, t[3], t[3]);
-
-  glm__memcpy(float, dest, t, sizeof(mat4));
-#endif
-}
+glm_mat4_mul(mat4 m1, mat4 m2, mat4 dest);
 
+/*!
+ * @brief translate existing transform matrix by v vector
+ *        and stores result in same matrix
+ *
+ * @param[in, out]  m  affine transfrom
+ * @param[in]       v  translate vector [x, y, z]
+ */
 CGLM_INLINE
 void
 glm_translate(mat4 m, vec3 v) {
 #if defined( __SSE__ ) || defined( __SSE2__ )
-  _mm_store_ps(m[3],
-               _mm_add_ps(_mm_add_ps(_mm_mul_ps(_mm_load_ps(m[0]),
-                                                _mm_set1_ps(v[0])),
-                                     _mm_mul_ps(_mm_load_ps(m[1]),
-                                                _mm_set1_ps(v[1]))),
-                          _mm_add_ps(_mm_mul_ps(_mm_load_ps(m[2]),
-                                                _mm_set1_ps(v[2])),
-                                     _mm_load_ps(m[3]))))
+  glmm_store(m[3],
+             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_load(m[0]),
+                                              _mm_set1_ps(v[0])),
+                                   _mm_mul_ps(glmm_load(m[1]),
+                                              _mm_set1_ps(v[1]))),
+                        _mm_add_ps(_mm_mul_ps(glmm_load(m[2]),
+                                              _mm_set1_ps(v[2])),
+                                   glmm_load(m[3]))))
   ;
 #else
   vec4 v1, v2, v3;
@@ -98,63 +77,110 @@ glm_translate(mat4 m, vec3 v) {
 #endif
 }
 
+/*!
+ * @brief translate existing transform matrix by v vector
+ *        and store result in dest
+ *
+ * source matrix will remain same
+ *
+ * @param[in]  m    affine transfrom
+ * @param[in]  v    translate vector [x, y, z]
+ * @param[out] dest translated matrix
+ */
 CGLM_INLINE
 void
-glm_translate_x(mat4 m, float to) {
+glm_translate_to(mat4 m, vec3 v, mat4 dest) {
+  glm_mat4_copy(m, dest);
+  glm_translate(dest, v);
+}
+
+/*!
+ * @brief translate existing transform matrix by x factor
+ *
+ * @param[in, out]  m  affine transfrom
+ * @param[in]       x  x factor
+ */
+CGLM_INLINE
+void
+glm_translate_x(mat4 m, float x) {
 #if defined( __SSE__ ) || defined( __SSE2__ )
-  _mm_store_ps(m[3],
-               _mm_add_ps(_mm_mul_ps(_mm_load_ps(m[0]),
-                                     _mm_set1_ps(to)),
-                          _mm_load_ps(m[3])))
+  glmm_store(m[3],
+             _mm_add_ps(_mm_mul_ps(glmm_load(m[0]),
+                                   _mm_set1_ps(x)),
+                        glmm_load(m[3])))
   ;
 #else
   vec4 v1;
-  glm_vec4_scale(m[0], to, v1);
+  glm_vec4_scale(m[0], x, v1);
   glm_vec4_add(v1, m[3], m[3]);
 #endif
 }
 
+/*!
+ * @brief translate existing transform matrix by y factor
+ *
+ * @param[in, out]  m  affine transfrom
+ * @param[in]       y  y factor
+ */
 CGLM_INLINE
 void
-glm_translate_y(mat4 m, float to) {
+glm_translate_y(mat4 m, float y) {
 #if defined( __SSE__ ) || defined( __SSE2__ )
-  _mm_store_ps(m[3],
-               _mm_add_ps(_mm_mul_ps(_mm_load_ps(m[1]),
-                                     _mm_set1_ps(to)),
-                          _mm_load_ps(m[3])))
+  glmm_store(m[3],
+             _mm_add_ps(_mm_mul_ps(glmm_load(m[1]),
+                                   _mm_set1_ps(y)),
+                        glmm_load(m[3])))
   ;
 #else
   vec4 v1;
-  glm_vec4_scale(m[1], to, v1);
+  glm_vec4_scale(m[1], y, v1);
   glm_vec4_add(v1, m[3], m[3]);
 #endif
 }
 
+/*!
+ * @brief translate existing transform matrix by z factor
+ *
+ * @param[in, out]  m  affine transfrom
+ * @param[in]       z  z factor
+ */
 CGLM_INLINE
 void
-glm_translate_z(mat4 m, float to) {
+glm_translate_z(mat4 m, float z) {
 #if defined( __SSE__ ) || defined( __SSE2__ )
-  _mm_store_ps(m[3],
-               _mm_add_ps(_mm_mul_ps(_mm_load_ps(m[2]),
-                                     _mm_set1_ps(to)),
-                          _mm_load_ps(m[3])))
+  glmm_store(m[3],
+             _mm_add_ps(_mm_mul_ps(glmm_load(m[2]),
+                                   _mm_set1_ps(z)),
+                        glmm_load(m[3])))
   ;
 #else
   vec4 v1;
-  glm_vec4_scale(m[2], to, v1);
+  glm_vec4_scale(m[2], z, v1);
   glm_vec4_add(v1, m[3], m[3]);
 #endif
 }
 
+/*!
+ * @brief creates NEW translate transform matrix by v vector
+ *
+ * @param[out]  m  affine transfrom
+ * @param[in]   v  translate vector [x, y, z]
+ */
 CGLM_INLINE
 void
 glm_translate_make(mat4 m, vec3 v) {
-  mat4 t = GLM_MAT4_IDENTITY_INIT;
-  glm_translate_to(t, v, m);
+  glm_mat4_identity(m);
+  glm_vec3_copy(v, m[3]);
 }
 
-/* scale */
-
+/*!
+ * @brief scale existing transform matrix by v vector
+ *        and store result in dest
+ *
+ * @param[in]  m    affine transfrom
+ * @param[in]  v    scale vector [x, y, z]
+ * @param[out] dest scaled matrix
+ */
 CGLM_INLINE
 void
 glm_scale_to(mat4 m, vec3 v, mat4 dest) {
@@ -165,173 +191,229 @@ glm_scale_to(mat4 m, vec3 v, mat4 dest) {
   glm_vec4_copy(m[3], dest[3]);
 }
 
+/*!
+ * @brief creates NEW scale matrix by v vector
+ *
+ * @param[out]  m  affine transfrom
+ * @param[in]   v  scale vector [x, y, z]
+ */
 CGLM_INLINE
 void
 glm_scale_make(mat4 m, vec3 v) {
-  mat4 t = GLM_MAT4_IDENTITY_INIT;
-  glm_scale_to(t, v, m);
+  glm_mat4_identity(m);
+  m[0][0] = v[0];
+  m[1][1] = v[1];
+  m[2][2] = v[2];
 }
 
+/*!
+ * @brief scales existing transform matrix by v vector
+ *        and stores result in same matrix
+ *
+ * @param[in, out]  m  affine transfrom
+ * @param[in]       v  scale vector [x, y, z]
+ */
 CGLM_INLINE
 void
 glm_scale(mat4 m, vec3 v) {
   glm_scale_to(m, v, m);
 }
 
+/*!
+ * @brief applies uniform scale to existing transform matrix v = [s, s, s]
+ *        and stores result in same matrix
+ *
+ * @param[in, out]  m  affine transfrom
+ * @param[in]       s  scale factor
+ */
 CGLM_INLINE
 void
-glm_scale1(mat4 m, float s) {
-  vec3 v = { s, s, s };
+glm_scale_uni(mat4 m, float s) {
+  CGLM_ALIGN(8) vec3 v = { s, s, s };
   glm_scale_to(m, v, m);
 }
 
+/*!
+ * @brief rotate existing transform matrix around X axis by angle
+ *        and store result in dest
+ *
+ * @param[in]   m      affine transfrom
+ * @param[in]   angle  angle (radians)
+ * @param[out]  dest   rotated matrix
+ */
 CGLM_INLINE
 void
-glm_rotate_x(mat4 m, float rad, mat4 dest) {
-  float cosVal;
-  float sinVal;
-  mat4  t = GLM_MAT4_IDENTITY_INIT;
+glm_rotate_x(mat4 m, float angle, mat4 dest) {
+  CGLM_ALIGN_MAT mat4 t = GLM_MAT4_IDENTITY_INIT;
+  float c, s;
 
-  cosVal = cosf(rad);
-  sinVal = sinf(rad);
+  c = cosf(angle);
+  s = sinf(angle);
 
-  t[1][1] =  cosVal;
-  t[1][2] =  sinVal;
-  t[2][1] = -sinVal;
-  t[2][2] =  cosVal;
+  t[1][1] =  c;
+  t[1][2] =  s;
+  t[2][1] = -s;
+  t[2][2] =  c;
 
-  glm_mat4_mul(m, t, dest);
+  glm_mul_rot(m, t, dest);
 }
 
+/*!
+ * @brief rotate existing transform matrix around Y axis by angle
+ *        and store result in dest
+ *
+ * @param[in]   m      affine transfrom
+ * @param[in]   angle  angle (radians)
+ * @param[out]  dest   rotated matrix
+ */
 CGLM_INLINE
 void
-glm_rotate_y(mat4 m, float rad, mat4 dest) {
-  float cosVal;
-  float sinVal;
-  mat4  t = GLM_MAT4_IDENTITY_INIT;
+glm_rotate_y(mat4 m, float angle, mat4 dest) {
+  CGLM_ALIGN_MAT mat4 t = GLM_MAT4_IDENTITY_INIT;
+  float c, s;
 
-  cosVal = cosf(rad);
-  sinVal = sinf(rad);
+  c = cosf(angle);
+  s = sinf(angle);
 
-  t[0][0] =  cosVal;
-  t[0][2] = -sinVal;
-  t[2][0] =  sinVal;
-  t[2][2] =  cosVal;
+  t[0][0] =  c;
+  t[0][2] = -s;
+  t[2][0] =  s;
+  t[2][2] =  c;
 
-  glm_mat4_mul(m, t, dest);
+  glm_mul_rot(m, t, dest);
 }
 
+/*!
+ * @brief rotate existing transform matrix around Z axis by angle
+ *        and store result in dest
+ *
+ * @param[in]   m      affine transfrom
+ * @param[in]   angle  angle (radians)
+ * @param[out]  dest   rotated matrix
+ */
 CGLM_INLINE
 void
-glm_rotate_z(mat4 m, float rad, mat4 dest) {
-  float cosVal;
-  float sinVal;
-  mat4  t = GLM_MAT4_IDENTITY_INIT;
+glm_rotate_z(mat4 m, float angle, mat4 dest) {
+  CGLM_ALIGN_MAT mat4 t = GLM_MAT4_IDENTITY_INIT;
+  float c, s;
 
-  cosVal = cosf(rad);
-  sinVal = sinf(rad);
+  c = cosf(angle);
+  s = sinf(angle);
 
-  t[0][0] =  cosVal;
-  t[0][1] =  sinVal;
-  t[1][0] = -sinVal;
-  t[1][1] =  cosVal;
+  t[0][0] =  c;
+  t[0][1] =  s;
+  t[1][0] = -s;
+  t[1][1] =  c;
 
-  glm_mat4_mul(m, t, dest);
+  glm_mul_rot(m, t, dest);
 }
 
+/*!
+ * @brief creates NEW rotation matrix by angle and axis
+ *
+ * axis will be normalized so you don't need to normalize it
+ *
+ * @param[out] m     affine transfrom
+ * @param[in]  angle angle (radians)
+ * @param[in]  axis  axis
+ */
 CGLM_INLINE
 void
-glm_rotate_ndc_make(mat4 m, float angle, vec3 axis_ndc) {
-  /* https://www.opengl.org/sdk/docs/man2/xhtml/glRotate.xml */
-
-  vec3 v, vs;
+glm_rotate_make(mat4 m, float angle, vec3 axis) {
+  CGLM_ALIGN(8) vec3 axisn, v, vs;
   float c;
 
   c = cosf(angle);
 
-  glm_vec_scale(axis_ndc, 1.0f - c, v);
-  glm_vec_scale(axis_ndc, sinf(angle), vs);
+  glm_vec3_normalize_to(axis, axisn);
+  glm_vec3_scale(axisn, 1.0f - c, v);
+  glm_vec3_scale(axisn, sinf(angle), vs);
 
-  glm_vec_scale(axis_ndc, v[0], m[0]);
-  glm_vec_scale(axis_ndc, v[1], m[1]);
-  glm_vec_scale(axis_ndc, v[2], m[2]);
+  glm_vec3_scale(axisn, v[0], m[0]);
+  glm_vec3_scale(axisn, v[1], m[1]);
+  glm_vec3_scale(axisn, v[2], m[2]);
 
-  m[0][0] += c;
-  m[0][1] += vs[2];
-  m[0][2] -= vs[1];
-
-  m[1][0] -= vs[2];
-  m[1][1] += c;
-  m[1][2] += vs[0];
-
-  m[2][0] += vs[1];
-  m[2][1] -= vs[0];
-  m[2][2] += c;
+  m[0][0] += c;       m[1][0] -= vs[2];   m[2][0] += vs[1];
+  m[0][1] += vs[2];   m[1][1] += c;       m[2][1] -= vs[0];
+  m[0][2] -= vs[1];   m[1][2] += vs[0];   m[2][2] += c;
 
   m[0][3] = m[1][3] = m[2][3] = m[3][0] = m[3][1] = m[3][2] = 0.0f;
   m[3][3] = 1.0f;
 }
 
-CGLM_INLINE
-void
-glm_rotate_make(mat4 m, float angle, vec3 axis) {
-  vec3 axis_ndc;
-
-  glm_vec_normalize_to(axis, axis_ndc);
-  glm_rotate_ndc_make(m, angle, axis_ndc);
-}
-
-CGLM_INLINE
-void
-glm_rotate_ndc(mat4 m, float angle, vec3 axis_ndc) {
-  mat4 rot, tmp;
-
-  glm_rotate_ndc_make(rot, angle, axis_ndc);
-
-  glm_vec4_scale(m[0], rot[0][0], tmp[1]);
-  glm_vec4_scale(m[1], rot[0][1], tmp[0]);
-  glm_vec4_add(tmp[1], tmp[0],    tmp[1]);
-  glm_vec4_scale(m[2], rot[0][2], tmp[0]);
-  glm_vec4_add(tmp[1], tmp[0],    tmp[1]);
-
-  glm_vec4_scale(m[0], rot[1][0], tmp[2]);
-  glm_vec4_scale(m[1], rot[1][1], tmp[0]);
-  glm_vec4_add(tmp[2], tmp[0],    tmp[2]);
-  glm_vec4_scale(m[2], rot[1][2], tmp[0]);
-  glm_vec4_add(tmp[2], tmp[0],    tmp[2]);
-
-  glm_vec4_scale(m[0], rot[2][0], tmp[3]);
-  glm_vec4_scale(m[1], rot[2][1], tmp[0]);
-  glm_vec4_add(tmp[3], tmp[0],    tmp[3]);
-  glm_vec4_scale(m[2], rot[2][2], tmp[0]);
-  glm_vec4_add(tmp[3], tmp[0],    tmp[3]);
-
-  glm_vec4_copy(tmp[1], m[0]);
-  glm_vec4_copy(tmp[2], m[1]);
-  glm_vec4_copy(tmp[3], m[2]);
-}
-
+/*!
+ * @brief rotate existing transform matrix around given axis by angle
+ *
+ * @param[in, out]  m      affine transfrom
+ * @param[in]       angle  angle (radians)
+ * @param[in]       axis   axis
+ */
 CGLM_INLINE
 void
 glm_rotate(mat4 m, float angle, vec3 axis) {
-  vec3 axis_ndc;
+  CGLM_ALIGN_MAT mat4 rot;
+  glm_rotate_make(rot, angle, axis);
+  glm_mul_rot(m, rot, m);
+}
 
-  glm_vec_normalize_to(axis, axis_ndc);
-  glm_rotate_ndc(m, angle, axis_ndc);
+/*!
+ * @brief rotate existing transform
+ *        around given axis by angle at given pivot point (rotation center)
+ *
+ * @param[in, out]  m      affine transfrom
+ * @param[in]       pivot  rotation center
+ * @param[in]       angle  angle (radians)
+ * @param[in]       axis   axis
+ */
+CGLM_INLINE
+void
+glm_rotate_at(mat4 m, vec3 pivot, float angle, vec3 axis) {
+  CGLM_ALIGN(8) vec3 pivotInv;
+
+  glm_vec3_negate_to(pivot, pivotInv);
+
+  glm_translate(m, pivot);
+  glm_rotate(m, angle, axis);
+  glm_translate(m, pivotInv);
+}
+
+/*!
+ * @brief creates NEW rotation matrix by angle and axis at given point
+ *
+ * this creates rotation matrix, it assumes you don't have a matrix
+ *
+ * this should work faster than glm_rotate_at because it reduces
+ * one glm_translate.
+ *
+ * @param[out] m      affine transfrom
+ * @param[in]  pivot  rotation center
+ * @param[in]  angle  angle (radians)
+ * @param[in]  axis   axis
+ */
+CGLM_INLINE
+void
+glm_rotate_atm(mat4 m, vec3 pivot, float angle, vec3 axis) {
+  CGLM_ALIGN(8) vec3 pivotInv;
+
+  glm_vec3_negate_to(pivot, pivotInv);
+
+  glm_translate_make(m, pivot);
+  glm_rotate(m, angle, axis);
+  glm_translate(m, pivotInv);
 }
 
 /*!
  * @brief decompose scale vector
  *
- * @param[in]  m     affine transform
- * @param[out] s     scale vector (Sx, Sy, Sz)
+ * @param[in]  m  affine transform
+ * @param[out] s  scale vector (Sx, Sy, Sz)
  */
 CGLM_INLINE
 void
 glm_decompose_scalev(mat4 m, vec3 s) {
-  s[0] = glm_vec_norm(m[0]);
-  s[1] = glm_vec_norm(m[1]);
-  s[2] = glm_vec_norm(m[2]);
+  s[0] = glm_vec3_norm(m[0]);
+  s[1] = glm_vec3_norm(m[1]);
+  s[2] = glm_vec3_norm(m[2]);
 }
 
 /*!
@@ -345,10 +427,9 @@ glm_decompose_scalev(mat4 m, vec3 s) {
 CGLM_INLINE
 bool
 glm_uniscaled(mat4 m) {
-  vec3 s;
+  CGLM_ALIGN(8) vec3 s;
   glm_decompose_scalev(m, s);
-
-  return glm_vec_eq_all(s);
+  return glm_vec3_eq_all(s);
 }
 
 /*!
@@ -362,17 +443,17 @@ glm_uniscaled(mat4 m) {
 CGLM_INLINE
 void
 glm_decompose_rs(mat4 m, mat4 r, vec3 s) {
-  vec4 t = {0.0f, 0.0f, 0.0f, 1.0f};
-  vec3 v;
+  CGLM_ALIGN(16) vec4 t = {0.0f, 0.0f, 0.0f, 1.0f};
+  CGLM_ALIGN(8)  vec3 v;
 
   glm_vec4_copy(m[0], r[0]);
   glm_vec4_copy(m[1], r[1]);
   glm_vec4_copy(m[2], r[2]);
   glm_vec4_copy(t,    r[3]);
 
-  s[0] = glm_vec_norm(m[0]);
-  s[1] = glm_vec_norm(m[1]);
-  s[2] = glm_vec_norm(m[2]);
+  s[0] = glm_vec3_norm(m[0]);
+  s[1] = glm_vec3_norm(m[1]);
+  s[2] = glm_vec3_norm(m[2]);
 
   glm_vec4_scale(r[0], 1.0f/s[0], r[0]);
   glm_vec4_scale(r[1], 1.0f/s[1], r[1]);
@@ -381,12 +462,12 @@ glm_decompose_rs(mat4 m, mat4 r, vec3 s) {
   /* Note from Apple Open Source (asume that the matrix is orthonormal):
      check for a coordinate system flip.  If the determinant
      is -1, then negate the matrix and the scaling factors. */
-  glm_vec_cross(m[0], m[1], v);
-  if (glm_vec_dot(v, m[2]) < 0.0f) {
-    glm_vec4_flipsign(r[0]);
-    glm_vec4_flipsign(r[1]);
-    glm_vec4_flipsign(r[2]);
-    glm_vec_flipsign(s);
+  glm_vec3_cross(m[0], m[1], v);
+  if (glm_vec3_dot(v, m[2]) < 0.0f) {
+    glm_vec4_negate(r[0]);
+    glm_vec4_negate(r[1]);
+    glm_vec4_negate(r[2]);
+    glm_vec3_negate(s);
   }
 }
 

include/cglm/applesimd.h

@@ -0,0 +1,95 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglm_applesimd_h
+#define cglm_applesimd_h
+#if defined(__APPLE__)                                                        \
+    && defined(SIMD_COMPILER_HAS_REQUIRED_FEATURES)                           \
+    && defined(SIMD_BASE)                                                     \
+    && defined(SIMD_TYPES)                                                    \
+    && defined(SIMD_VECTOR_TYPES)
+
+#include "common.h"
+
+/*!
+* @brief converts mat4 to Apple's simd type simd_float4x4
+* @return simd_float4x4
+*/
+CGLM_INLINE
+simd_float4x4
+glm_mat4_applesimd(mat4 m) {
+  simd_float4x4 t;
+  
+  t.columns[0][0] = m[0][0];
+  t.columns[0][1] = m[0][1];
+  t.columns[0][2] = m[0][2];
+  t.columns[0][3] = m[0][3];
+
+  t.columns[1][0] = m[1][0];
+  t.columns[1][1] = m[1][1];
+  t.columns[1][2] = m[1][2];
+  t.columns[1][3] = m[1][3];
+
+  t.columns[2][0] = m[2][0];
+  t.columns[2][1] = m[2][1];
+  t.columns[2][2] = m[2][2];
+  t.columns[2][3] = m[2][3];
+
+  t.columns[3][0] = m[3][0];
+  t.columns[3][1] = m[3][1];
+  t.columns[3][2] = m[3][2];
+  t.columns[3][3] = m[3][3];
+
+  return t;
+}
+
+/*!
+* @brief converts mat3 to Apple's simd type simd_float3x3
+* @return simd_float3x3
+*/
+CGLM_INLINE
+simd_float3x3
+glm_mat3_applesimd(mat3 m) {
+  simd_float3x3 t;
+  
+  t.columns[0][0] = m[0][0];
+  t.columns[0][1] = m[0][1];
+  t.columns[0][2] = m[0][2];
+
+  t.columns[1][0] = m[1][0];
+  t.columns[1][1] = m[1][1];
+  t.columns[1][2] = m[1][2];
+
+  t.columns[2][0] = m[2][0];
+  t.columns[2][1] = m[2][1];
+  t.columns[2][2] = m[2][2];
+
+  return t;
+}
+
+/*!
+* @brief converts vec4 to Apple's simd type simd_float4
+* @return simd_float4
+*/
+CGLM_INLINE
+simd_float4
+glm_vec4_applesimd(vec4 v) {
+  return (simd_float4){v[0], v[1], v[2], v[3]};
+}
+
+/*!
+* @brief converts vec3 to Apple's simd type simd_float3
+* @return v
+*/
+CGLM_INLINE
+simd_float3
+glm_vec3_applesimd(vec3 v) {
+  return (simd_float3){v[0], v[1], v[2]};
+}
+
+#endif
+#endif /* cglm_applesimd_h */

include/cglm/bezier.h

@@ -0,0 +1,154 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglm_bezier_h
+#define cglm_bezier_h
+
+#include "common.h"
+
+#define GLM_BEZIER_MAT_INIT  {{-1.0f,  3.0f, -3.0f,  1.0f},                   \
+                              { 3.0f, -6.0f,  3.0f,  0.0f},                   \
+                              {-3.0f,  3.0f,  0.0f,  0.0f},                   \
+                              { 1.0f,  0.0f,  0.0f,  0.0f}}
+#define GLM_HERMITE_MAT_INIT {{ 2.0f, -3.0f,  0.0f,  1.0f},                   \
+                              {-2.0f,  3.0f,  0.0f,  0.0f},                   \
+                              { 1.0f, -2.0f,  1.0f,  0.0f},                   \
+                              { 1.0f, -1.0f,  0.0f,  0.0f}}
+/* for C only */
+#define GLM_BEZIER_MAT  ((mat4)GLM_BEZIER_MAT_INIT)
+#define GLM_HERMITE_MAT ((mat4)GLM_HERMITE_MAT_INIT)
+
+#define CGLM_DECASTEL_EPS   1e-9f
+#define CGLM_DECASTEL_MAX   1000.0f
+#define CGLM_DECASTEL_SMALL 1e-20f
+
+/*!
+ * @brief cubic bezier interpolation
+ *
+ * Formula:
+ *  B(s) = P0*(1-s)^3 + 3*C0*s*(1-s)^2 + 3*C1*s^2*(1-s) + P1*s^3
+ *
+ * similar result using matrix:
+ *  B(s) = glm_smc(t, GLM_BEZIER_MAT, (vec4){p0, c0, c1, p1})
+ *
+ * glm_eq(glm_smc(...), glm_bezier(...)) should return TRUE
+ *
+ * @param[in]  s    parameter between 0 and 1
+ * @param[in]  p0   begin point
+ * @param[in]  c0   control point 1
+ * @param[in]  c1   control point 2
+ * @param[in]  p1   end point
+ *
+ * @return B(s)
+ */
+CGLM_INLINE
+float
+glm_bezier(float s, float p0, float c0, float c1, float p1) {
+  float x, xx, ss, xs3, a;
+
+  x   = 1.0f - s;
+  xx  = x * x;
+  ss  = s * s;
+  xs3 = (s - ss) * 3.0f;
+  a   = p0 * xx + c0 * xs3;
+
+  return a + s * (c1 * xs3 + p1 * ss - a);
+}
+
+/*!
+ * @brief cubic hermite interpolation
+ *
+ * Formula:
+ *  H(s) = P0*(2*s^3 - 3*s^2 + 1) + T0*(s^3 - 2*s^2 + s)
+ *            + P1*(-2*s^3 + 3*s^2) + T1*(s^3 - s^2)
+ *
+ * similar result using matrix:
+ *  H(s) = glm_smc(t, GLM_HERMITE_MAT, (vec4){p0, p1, c0, c1})
+ *
+ * glm_eq(glm_smc(...), glm_hermite(...)) should return TRUE
+ *
+ * @param[in]  s    parameter between 0 and 1
+ * @param[in]  p0   begin point
+ * @param[in]  t0   tangent 1
+ * @param[in]  t1   tangent 2
+ * @param[in]  p1   end point
+ *
+ * @return H(s)
+ */
+CGLM_INLINE
+float
+glm_hermite(float s, float p0, float t0, float t1, float p1) {
+  float ss, d, a, b, c, e, f;
+
+  ss = s  * s;
+  a  = ss + ss;
+  c  = a  + ss;
+  b  = a  * s;
+  d  = s  * ss;
+  f  = d  - ss;
+  e  = b  - c;
+
+  return p0 * (e + 1.0f) + t0 * (f - ss + s) + t1 * f - p1 * e;
+}
+
+/*!
+ * @brief iterative way to solve cubic equation
+ *
+ * @param[in]  prm  parameter between 0 and 1
+ * @param[in]  p0   begin point
+ * @param[in]  c0   control point 1
+ * @param[in]  c1   control point 2
+ * @param[in]  p1   end point
+ *
+ * @return parameter to use in cubic equation
+ */
+CGLM_INLINE
+float
+glm_decasteljau(float prm, float p0, float c0, float c1, float p1) {
+  float u, v, a, b, c, d, e, f;
+  int   i;
+
+  if (prm - p0 < CGLM_DECASTEL_SMALL)
+    return 0.0f;
+
+  if (p1 - prm < CGLM_DECASTEL_SMALL)
+    return 1.0f;
+
+  u  = 0.0f;
+  v  = 1.0f;
+
+  for (i = 0; i < CGLM_DECASTEL_MAX; i++) {
+    /* de Casteljau Subdivision */
+    a  = (p0 + c0) * 0.5f;
+    b  = (c0 + c1) * 0.5f;
+    c  = (c1 + p1) * 0.5f;
+    d  = (a  + b)  * 0.5f;
+    e  = (b  + c)  * 0.5f;
+    f  = (d  + e)  * 0.5f; /* this one is on the curve! */
+
+    /* The curve point is close enough to our wanted t */
+    if (fabsf(f - prm) < CGLM_DECASTEL_EPS)
+      return glm_clamp_zo((u  + v) * 0.5f);
+
+    /* dichotomy */
+    if (f < prm) {
+      p0 = f;
+      c0 = e;
+      c1 = c;
+      u  = (u  + v) * 0.5f;
+    } else {
+      c0 = a;
+      c1 = d;
+      p1 = f;
+      v  = (u  + v) * 0.5f;
+    }
+  }
+
+  return glm_clamp_zo((u  + v) * 0.5f);
+}
+
+#endif /* cglm_bezier_h */

include/cglm/box.h

@@ -0,0 +1,279 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglm_box_h
+#define cglm_box_h
+
+#include "common.h"
+#include "vec3.h"
+#include "vec4.h"
+#include "util.h"
+
+/*!
+ * @brief apply transform to Axis-Aligned Bounding Box
+ *
+ * @param[in]  box  bounding box
+ * @param[in]  m    transform matrix
+ * @param[out] dest transformed bounding box
+ */
+CGLM_INLINE
+void
+glm_aabb_transform(vec3 box[2], mat4 m, vec3 dest[2]) {
+  vec3 v[2], xa, xb, ya, yb, za, zb;
+
+  glm_vec3_scale(m[0], box[0][0], xa);
+  glm_vec3_scale(m[0], box[1][0], xb);
+
+  glm_vec3_scale(m[1], box[0][1], ya);
+  glm_vec3_scale(m[1], box[1][1], yb);
+
+  glm_vec3_scale(m[2], box[0][2], za);
+  glm_vec3_scale(m[2], box[1][2], zb);
+
+  /* translation + min(xa, xb) + min(ya, yb) + min(za, zb) */
+  glm_vec3(m[3], v[0]);
+  glm_vec3_minadd(xa, xb, v[0]);
+  glm_vec3_minadd(ya, yb, v[0]);
+  glm_vec3_minadd(za, zb, v[0]);
+
+  /* translation + max(xa, xb) + max(ya, yb) + max(za, zb) */
+  glm_vec3(m[3], v[1]);
+  glm_vec3_maxadd(xa, xb, v[1]);
+  glm_vec3_maxadd(ya, yb, v[1]);
+  glm_vec3_maxadd(za, zb, v[1]);
+
+  glm_vec3_copy(v[0], dest[0]);
+  glm_vec3_copy(v[1], dest[1]);
+}
+
+/*!
+ * @brief merges two AABB bounding box and creates new one
+ *
+ * two box must be in same space, if one of box is in different space then
+ * you should consider to convert it's space by glm_box_space
+ *
+ * @param[in]  box1 bounding box 1
+ * @param[in]  box2 bounding box 2
+ * @param[out] dest merged bounding box
+ */
+CGLM_INLINE
+void
+glm_aabb_merge(vec3 box1[2], vec3 box2[2], vec3 dest[2]) {
+  dest[0][0] = glm_min(box1[0][0], box2[0][0]);
+  dest[0][1] = glm_min(box1[0][1], box2[0][1]);
+  dest[0][2] = glm_min(box1[0][2], box2[0][2]);
+
+  dest[1][0] = glm_max(box1[1][0], box2[1][0]);
+  dest[1][1] = glm_max(box1[1][1], box2[1][1]);
+  dest[1][2] = glm_max(box1[1][2], box2[1][2]);
+}
+
+/*!
+ * @brief crops a bounding box with another one.
+ *
+ * this could be useful for gettng a bbox which fits with view frustum and
+ * object bounding boxes. In this case you crop view frustum box with objects
+ * box
+ *
+ * @param[in]  box     bounding box 1
+ * @param[in]  cropBox crop box
+ * @param[out] dest    cropped bounding box
+ */
+CGLM_INLINE
+void
+glm_aabb_crop(vec3 box[2], vec3 cropBox[2], vec3 dest[2]) {
+  dest[0][0] = glm_max(box[0][0], cropBox[0][0]);
+  dest[0][1] = glm_max(box[0][1], cropBox[0][1]);
+  dest[0][2] = glm_max(box[0][2], cropBox[0][2]);
+
+  dest[1][0] = glm_min(box[1][0], cropBox[1][0]);
+  dest[1][1] = glm_min(box[1][1], cropBox[1][1]);
+  dest[1][2] = glm_min(box[1][2], cropBox[1][2]);
+}
+
+/*!
+ * @brief crops a bounding box with another one.
+ *
+ * this could be useful for gettng a bbox which fits with view frustum and
+ * object bounding boxes. In this case you crop view frustum box with objects
+ * box
+ *
+ * @param[in]  box      bounding box
+ * @param[in]  cropBox  crop box
+ * @param[in]  clampBox miniumum box
+ * @param[out] dest     cropped bounding box
+ */
+CGLM_INLINE
+void
+glm_aabb_crop_until(vec3 box[2],
+                    vec3 cropBox[2],
+                    vec3 clampBox[2],
+                    vec3 dest[2]) {
+  glm_aabb_crop(box, cropBox, dest);
+  glm_aabb_merge(clampBox, dest, dest);
+}
+
+/*!
+ * @brief check if AABB intersects with frustum planes
+ *
+ * this could be useful for frustum culling using AABB.
+ *
+ * OPTIMIZATION HINT:
+ *  if planes order is similar to LEFT, RIGHT, BOTTOM, TOP, NEAR, FAR
+ *  then this method should run even faster because it would only use two
+ *  planes if object is not inside the two planes
+ *  fortunately cglm extracts planes as this order! just pass what you got!
+ *
+ * @param[in]  box     bounding box
+ * @param[in]  planes  frustum planes
+ */
+CGLM_INLINE
+bool
+glm_aabb_frustum(vec3 box[2], vec4 planes[6]) {
+  float *p, dp;
+  int    i;
+
+  for (i = 0; i < 6; i++) {
+    p  = planes[i];
+    dp = p[0] * box[p[0] > 0.0f][0]
+       + p[1] * box[p[1] > 0.0f][1]
+       + p[2] * box[p[2] > 0.0f][2];
+
+    if (dp < -p[3])
+      return false;
+  }
+
+  return true;
+}
+
+/*!
+ * @brief invalidate AABB min and max values
+ *
+ * @param[in, out]  box bounding box
+ */
+CGLM_INLINE
+void
+glm_aabb_invalidate(vec3 box[2]) {
+  glm_vec3_broadcast(FLT_MAX,  box[0]);
+  glm_vec3_broadcast(-FLT_MAX, box[1]);
+}
+
+/*!
+ * @brief check if AABB is valid or not
+ *
+ * @param[in]  box bounding box
+ */
+CGLM_INLINE
+bool
+glm_aabb_isvalid(vec3 box[2]) {
+  return glm_vec3_max(box[0]) != FLT_MAX
+         && glm_vec3_min(box[1]) != -FLT_MAX;
+}
+
+/*!
+ * @brief distance between of min and max
+ *
+ * @param[in]  box bounding box
+ */
+CGLM_INLINE
+float
+glm_aabb_size(vec3 box[2]) {
+  return glm_vec3_distance(box[0], box[1]);
+}
+
+/*!
+ * @brief radius of sphere which surrounds AABB
+ *
+ * @param[in]  box bounding box
+ */
+CGLM_INLINE
+float
+glm_aabb_radius(vec3 box[2]) {
+  return glm_aabb_size(box) * 0.5f;
+}
+
+/*!
+ * @brief computes center point of AABB
+ *
+ * @param[in]   box  bounding box
+ * @param[out]  dest center of bounding box
+ */
+CGLM_INLINE
+void
+glm_aabb_center(vec3 box[2], vec3 dest) {
+  glm_vec3_center(box[0], box[1], dest);
+}
+
+/*!
+ * @brief check if two AABB intersects
+ *
+ * @param[in]   box    bounding box
+ * @param[in]   other  other bounding box
+ */
+CGLM_INLINE
+bool
+glm_aabb_aabb(vec3 box[2], vec3 other[2]) {
+  return (box[0][0] <= other[1][0] && box[1][0] >= other[0][0])
+      && (box[0][1] <= other[1][1] && box[1][1] >= other[0][1])
+      && (box[0][2] <= other[1][2] && box[1][2] >= other[0][2]);
+}
+
+/*!
+ * @brief check if AABB intersects with sphere
+ *
+ * https://github.com/erich666/GraphicsGems/blob/master/gems/BoxSphere.c
+ * Solid Box - Solid Sphere test.
+ *
+ * @param[in]   box    solid bounding box
+ * @param[in]   s      solid sphere
+ */
+CGLM_INLINE
+bool
+glm_aabb_sphere(vec3 box[2], vec4 s) {
+  float dmin;
+  int   a, b, c;
+
+  a = s[0] >= box[0][0];
+  b = s[1] >= box[0][1];
+  c = s[2] >= box[0][2];
+
+  dmin  = glm_pow2(s[0] - box[a][0])
+        + glm_pow2(s[1] - box[b][1])
+        + glm_pow2(s[2] - box[c][2]);
+
+  return dmin <= glm_pow2(s[3]);
+}
+
+/*!
+ * @brief check if point is inside of AABB
+ *
+ * @param[in]   box    bounding box
+ * @param[in]   point  point
+ */
+CGLM_INLINE
+bool
+glm_aabb_point(vec3 box[2], vec3 point) {
+  return (point[0] >= box[0][0] && point[0] <= box[1][0])
+      && (point[1] >= box[0][1] && point[1] <= box[1][1])
+      && (point[2] >= box[0][2] && point[2] <= box[1][2]);
+}
+
+/*!
+ * @brief check if AABB contains other AABB
+ *
+ * @param[in]   box    bounding box
+ * @param[in]   other  other bounding box
+ */
+CGLM_INLINE
+bool
+glm_aabb_contains(vec3 box[2], vec3 other[2]) {
+  return (box[0][0] <= other[0][0] && box[1][0] >= other[1][0])
+      && (box[0][1] <= other[0][1] && box[1][1] >= other[1][1])
+      && (box[0][2] <= other[0][2] && box[1][2] >= other[1][2]);
+}
+
+#endif /* cglm_box_h */

include/cglm/call.h

@@ -12,15 +12,25 @@ extern "C" {
 #endif
 
 #include "cglm.h"
+#include "call/vec2.h"
 #include "call/vec3.h"
 #include "call/vec4.h"
-#include "call/mat4.h"
+#include "call/mat2.h"
 #include "call/mat3.h"
+#include "call/mat4.h"
 #include "call/affine.h"
 #include "call/cam.h"
 #include "call/quat.h"
 #include "call/euler.h"
+#include "call/plane.h"
+#include "call/frustum.h"
+#include "call/box.h"
 #include "call/io.h"
+#include "call/project.h"
+#include "call/sphere.h"
+#include "call/ease.h"
+#include "call/curve.h"
+#include "call/bezier.h"
 
 #ifdef __cplusplus
 }

include/cglm/call/affine.h

@@ -13,6 +13,10 @@ extern "C" {
 
 #include "../cglm.h"
 
+CGLM_EXPORT
+void
+glmc_translate_make(mat4 m, vec3 v);
+
 CGLM_EXPORT
 void
 glmc_translate_to(mat4 m, vec3 v, mat4 dest);
@@ -33,6 +37,10 @@ CGLM_EXPORT
 void
 glmc_translate_z(mat4 m, float to);
 
+CGLM_EXPORT
+void
+glmc_scale_make(mat4 m, vec3 v);
+
 CGLM_EXPORT
 void
 glmc_scale_to(mat4 m, vec3 v, mat4 dest);
@@ -43,7 +51,7 @@ glmc_scale(mat4 m, vec3 v);
 
 CGLM_EXPORT
 void
-glmc_scale1(mat4 m, float s);
+glmc_scale_uni(mat4 m, float s);
 
 CGLM_EXPORT
 void
@@ -57,26 +65,30 @@ CGLM_EXPORT
 void
 glmc_rotate_z(mat4 m, float rad, mat4 dest);
 
-CGLM_EXPORT
-void
-glmc_rotate_ndc_make(mat4 m, float angle, vec3 axis_ndc);
-
 CGLM_EXPORT
 void
 glmc_rotate_make(mat4 m, float angle, vec3 axis);
 
-CGLM_EXPORT
-void
-glmc_rotate_ndc(mat4 m, float angle, vec3 axis_ndc);
-
 CGLM_EXPORT
 void
 glmc_rotate(mat4 m, float angle, vec3 axis);
 
+CGLM_EXPORT
+void
+glmc_rotate_at(mat4 m, vec3 pivot, float angle, vec3 axis);
+
+CGLM_EXPORT
+void
+glmc_rotate_atm(mat4 m, vec3 pivot, float angle, vec3 axis);
+
 CGLM_EXPORT
 void
 glmc_decompose_scalev(mat4 m, vec3 s);
 
+CGLM_EXPORT
+bool
+glmc_uniscaled(mat4 m);
+
 CGLM_EXPORT
 void
 glmc_decompose_rs(mat4 m, mat4 r, vec3 s);
@@ -85,6 +97,20 @@ CGLM_EXPORT
 void
 glmc_decompose(mat4 m, vec4 t, mat4 r, vec3 s);
 
+/* affine-mat */
+
+CGLM_EXPORT
+void
+glmc_mul(mat4 m1, mat4 m2, mat4 dest);
+
+CGLM_EXPORT
+void
+glmc_mul_rot(mat4 m1, mat4 m2, mat4 dest);
+
+CGLM_EXPORT
+void
+glmc_inv_tr(mat4 mat);
+
 #ifdef __cplusplus
 }
 #endif

include/cglm/call/bezier.h

@@ -0,0 +1,31 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglmc_bezier_h
+#define cglmc_bezier_h
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "../cglm.h"
+
+CGLM_EXPORT
+float
+glmc_bezier(float s, float p0, float c0, float c1, float p1);
+
+CGLM_EXPORT
+float
+glmc_hermite(float s, float p0, float t0, float t1, float p1);
+
+CGLM_EXPORT
+float
+glmc_decasteljau(float prm, float p0, float c0, float c1, float p1);
+
+#ifdef __cplusplus
+}
+#endif
+#endif /* cglmc_bezier_h */

include/cglm/call/box.h

@@ -0,0 +1,79 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglmc_box_h
+#define cglmc_box_h
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "../cglm.h"
+
+CGLM_EXPORT
+void
+glmc_aabb_transform(vec3 box[2], mat4 m, vec3 dest[2]);
+
+CGLM_EXPORT
+void
+glmc_aabb_merge(vec3 box1[2], vec3 box2[2], vec3 dest[2]);
+
+CGLM_EXPORT
+void
+glmc_aabb_crop(vec3 box[2], vec3 cropBox[2], vec3 dest[2]);
+
+CGLM_EXPORT
+void
+glmc_aabb_crop_until(vec3 box[2],
+                     vec3 cropBox[2],
+                     vec3 clampBox[2],
+                     vec3 dest[2]);
+
+CGLM_EXPORT
+bool
+glmc_aabb_frustum(vec3 box[2], vec4 planes[6]);
+
+CGLM_EXPORT
+void
+glmc_aabb_invalidate(vec3 box[2]);
+
+CGLM_EXPORT
+bool
+glmc_aabb_isvalid(vec3 box[2]);
+
+CGLM_EXPORT
+float
+glmc_aabb_size(vec3 box[2]);
+
+CGLM_EXPORT
+float
+glmc_aabb_radius(vec3 box[2]);
+
+CGLM_EXPORT
+void
+glmc_aabb_center(vec3 box[2], vec3 dest);
+
+CGLM_EXPORT
+bool
+glmc_aabb_aabb(vec3 box[2], vec3 other[2]);
+
+CGLM_EXPORT
+bool
+glmc_aabb_point(vec3 box[2], vec3 point);
+
+CGLM_EXPORT
+bool
+glmc_aabb_contains(vec3 box[2], vec3 other[2]);
+
+CGLM_EXPORT
+bool
+glmc_aabb_sphere(vec3 box[2], vec4 s);
+
+#ifdef __cplusplus
+}
+#endif
+#endif /* cglmc_box_h */
+

include/cglm/call/cam.h

@@ -33,6 +33,26 @@ glmc_ortho(float left,
            float farVal,
            mat4 dest);
 
+CGLM_EXPORT
+void
+glmc_ortho_aabb(vec3 box[2], mat4 dest);
+
+CGLM_EXPORT
+void
+glmc_ortho_aabb_p(vec3 box[2], float padding, mat4 dest);
+
+CGLM_EXPORT
+void
+glmc_ortho_aabb_pz(vec3 box[2], float padding, mat4 dest);
+
+CGLM_EXPORT
+void
+glmc_ortho_default(float aspect, mat4 dest);
+
+CGLM_EXPORT
+void
+glmc_ortho_default_s(float aspect, float size, mat4 dest);
+
 CGLM_EXPORT
 void
 glmc_perspective(float fovy,
@@ -43,10 +63,79 @@ glmc_perspective(float fovy,
 
 CGLM_EXPORT
 void
-glmc_lookat(vec3 eye,
-            vec3 center,
-            vec3 up,
-            mat4 dest);
+glmc_persp_move_far(mat4 proj, float deltaFar);
+
+CGLM_EXPORT
+void
+glmc_perspective_default(float aspect, mat4 dest);
+
+CGLM_EXPORT
+void
+glmc_perspective_resize(float aspect, mat4 proj);
+
+CGLM_EXPORT
+void
+glmc_lookat(vec3 eye, vec3 center, vec3 up, mat4 dest);
+
+CGLM_EXPORT
+void
+glmc_look(vec3 eye, vec3 dir, vec3 up, mat4 dest);
+
+CGLM_EXPORT
+void
+glmc_look_anyup(vec3 eye, vec3 dir, mat4 dest);
+
+CGLM_EXPORT
+void
+glmc_persp_decomp(mat4 proj,
+                  float * __restrict nearVal,
+                  float * __restrict farVal,
+                  float * __restrict top,
+                  float * __restrict bottom,
+                  float * __restrict left,
+                  float * __restrict right);
+
+CGLM_EXPORT
+void
+glmc_persp_decompv(mat4 proj, float dest[6]);
+
+CGLM_EXPORT
+void
+glmc_persp_decomp_x(mat4 proj,
+                    float * __restrict left,
+                    float * __restrict right);
+
+CGLM_EXPORT
+void
+glmc_persp_decomp_y(mat4 proj,
+                    float * __restrict top,
+                    float * __restrict bottom);
+
+CGLM_EXPORT
+void
+glmc_persp_decomp_z(mat4 proj,
+                    float * __restrict nearVal,
+                    float * __restrict farVal);
+
+CGLM_EXPORT
+void
+glmc_persp_decomp_far(mat4 proj, float * __restrict farVal);
+
+CGLM_EXPORT
+void
+glmc_persp_decomp_near(mat4 proj, float * __restrict nearVal);
+
+CGLM_EXPORT
+float
+glmc_persp_fovy(mat4 proj);
+
+CGLM_EXPORT
+float
+glmc_persp_aspect(mat4 proj);
+
+CGLM_EXPORT
+void
+glmc_persp_sizes(mat4 proj, float fovy, vec4 dest);
 
 #ifdef __cplusplus
 }

include/cglm/call/curve.h

@@ -0,0 +1,23 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglmc_curve_h
+#define cglmc_curve_h
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "../cglm.h"
+
+CGLM_EXPORT
+float
+glmc_smc(float s, mat4 m, vec4 c);
+
+#ifdef __cplusplus
+}
+#endif
+#endif /* cglmc_curve_h */

include/cglm/call/ease.h

@@ -0,0 +1,143 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglmc_ease_h
+#define cglmc_ease_h
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "../cglm.h"
+
+CGLM_EXPORT
+float
+glmc_ease_linear(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_sine_in(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_sine_out(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_sine_inout(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_quad_in(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_quad_out(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_quad_inout(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_cubic_in(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_cubic_out(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_cubic_inout(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_quart_in(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_quart_out(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_quart_inout(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_quint_in(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_quint_out(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_quint_inout(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_exp_in(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_exp_out(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_exp_inout(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_circ_in(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_circ_out(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_circ_inout(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_back_in(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_back_out(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_back_inout(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_elast_in(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_elast_out(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_elast_inout(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_bounce_out(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_bounce_in(float t);
+
+CGLM_EXPORT
+float
+glmc_ease_bounce_inout(float t);
+
+#ifdef __cplusplus
+}
+#endif
+#endif /* cglmc_ease_h */

include/cglm/call/euler.h

@@ -21,6 +21,10 @@ CGLM_EXPORT
 void
 glmc_euler(vec3 angles, mat4 dest);
 
+CGLM_EXPORT
+void
+glmc_euler_xyz(vec3 angles,  mat4 dest);
+
 CGLM_EXPORT
 void
 glmc_euler_zyx(vec3 angles,  mat4 dest);
@@ -43,7 +47,7 @@ glmc_euler_yxz(vec3 angles, mat4 dest);
 
 CGLM_EXPORT
 void
-glmc_euler_by_order(vec3 angles, glm_euler_sq axis, mat4 dest);
+glmc_euler_by_order(vec3 angles, glm_euler_seq axis, mat4 dest);
 
 #ifdef __cplusplus
 }

include/cglm/call/frustum.h

@@ -0,0 +1,41 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglmc_frustum_h
+#define cglmc_frustum_h
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "../cglm.h"
+
+CGLM_EXPORT
+void
+glmc_frustum_planes(mat4 m, vec4 dest[6]);
+
+CGLM_EXPORT
+void
+glmc_frustum_corners(mat4 invMat, vec4 dest[8]);
+
+CGLM_EXPORT
+void
+glmc_frustum_center(vec4 corners[8], vec4 dest);
+
+CGLM_EXPORT
+void
+glmc_frustum_box(vec4 corners[8], mat4 m, vec3 box[2]);
+
+CGLM_EXPORT
+void
+glmc_frustum_corners_at(vec4  corners[8],
+                        float splitDist,
+                        float farDist,
+                        vec4  planeCorners[4]);
+#ifdef __cplusplus
+}
+#endif
+#endif /* cglmc_frustum_h */

include/cglm/call/mat2.h

@@ -0,0 +1,79 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglmc_mat2_h
+#define cglmc_mat2_h
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "../cglm.h"
+
+CGLM_EXPORT
+void
+glmc_mat2_copy(mat2 mat, mat2 dest);
+
+CGLM_EXPORT
+void
+glmc_mat2_identity(mat2 mat);
+
+CGLM_EXPORT
+void
+glmc_mat2_identity_array(mat2 * __restrict mat, size_t count);
+
+CGLM_EXPORT
+void
+glmc_mat2_zero(mat2 mat);
+
+CGLM_EXPORT
+void
+glmc_mat2_mul(mat2 m1, mat2 m2, mat2 dest);
+
+CGLM_EXPORT
+void
+glmc_mat2_transpose_to(mat2 m, mat2 dest);
+
+CGLM_EXPORT
+void
+glmc_mat2_transpose(mat2 m);
+
+CGLM_EXPORT
+void
+glmc_mat2_mulv(mat2 m, vec2 v, vec2 dest);
+
+CGLM_EXPORT
+float
+glmc_mat2_trace(mat2 m);
+
+CGLM_EXPORT
+void
+glmc_mat2_scale(mat2 m, float s);
+
+CGLM_EXPORT
+float
+glmc_mat2_det(mat2 mat);
+
+CGLM_EXPORT
+void
+glmc_mat2_inv(mat2 mat, mat2 dest);
+
+CGLM_EXPORT
+void
+glmc_mat2_swap_col(mat2 mat, int col1, int col2);
+
+CGLM_EXPORT
+void
+glmc_mat2_swap_row(mat2 mat, int row1, int row2);
+
+CGLM_EXPORT
+float
+glmc_mat2_rmc(vec2 r, mat2 m, vec2 c);
+
+#ifdef __cplusplus
+}
+#endif
+#endif /* cglmc_mat2_h */

include/cglm/call/mat3.h

@@ -24,6 +24,14 @@ CGLM_EXPORT
 void
 glmc_mat3_identity(mat3 mat);
 
+CGLM_EXPORT
+void
+glmc_mat3_zero(mat3 mat);
+
+CGLM_EXPORT
+void
+glmc_mat3_identity_array(mat3 * __restrict mat, size_t count);
+
 CGLM_EXPORT
 void
 glmc_mat3_mul(mat3 m1, mat3 m2, mat3 dest);
@@ -40,6 +48,14 @@ CGLM_EXPORT
 void
 glmc_mat3_mulv(mat3 m, vec3 v, vec3 dest);
 
+CGLM_EXPORT
+float
+glmc_mat3_trace(mat3 m);
+
+CGLM_EXPORT
+void
+glmc_mat3_quat(mat3 m, versor dest);
+
 CGLM_EXPORT
 void
 glmc_mat3_scale(mat3 m, float s);
@@ -60,6 +76,10 @@ CGLM_EXPORT
 void
 glmc_mat3_swap_row(mat3 mat, int row1, int row2);
 
+CGLM_EXPORT
+float
+glmc_mat3_rmc(vec3 r, mat3 m, vec3 c);
+
 #ifdef __cplusplus
 }
 #endif

include/cglm/call/mat4.h

@@ -29,6 +29,14 @@ CGLM_EXPORT
 void
 glmc_mat4_identity(mat4 mat);
 
+CGLM_EXPORT
+void
+glmc_mat4_identity_array(mat4 * __restrict mat, size_t count);
+
+CGLM_EXPORT
+void
+glmc_mat4_zero(mat4 mat);
+
 CGLM_EXPORT
 void
 glmc_mat4_pick3(mat4 mat, mat3 dest);
@@ -47,12 +55,28 @@ glmc_mat4_mul(mat4 m1, mat4 m2, mat4 dest);
 
 CGLM_EXPORT
 void
-glmc_mat4_mulN(mat4 * __restrict matrices[], int len, mat4 dest);
+glmc_mat4_mulN(mat4 * __restrict matrices[], uint32_t len, mat4 dest);
 
 CGLM_EXPORT
 void
 glmc_mat4_mulv(mat4 m, vec4 v, vec4 dest);
 
+CGLM_EXPORT
+void
+glmc_mat4_mulv3(mat4 m, vec3 v, float last, vec3 dest);
+
+CGLM_EXPORT
+float
+glmc_mat4_trace(mat4 m);
+
+CGLM_EXPORT
+float
+glmc_mat4_trace3(mat4 m);
+
+CGLM_EXPORT
+void
+glmc_mat4_quat(mat4 m, versor dest);
+
 CGLM_EXPORT
 void
 glmc_mat4_transpose_to(mat4 m, mat4 dest);
@@ -81,6 +105,10 @@ CGLM_EXPORT
 void
 glmc_mat4_inv_precise(mat4 mat, mat4 dest);
 
+CGLM_EXPORT
+void
+glmc_mat4_inv_fast(mat4 mat, mat4 dest);
+
 CGLM_EXPORT
 void
 glmc_mat4_swap_col(mat4 mat, int col1, int col2);
@@ -89,6 +117,10 @@ CGLM_EXPORT
 void
 glmc_mat4_swap_row(mat4 mat, int row1, int row2);
 
+CGLM_EXPORT
+float
+glmc_mat4_rmc(vec4 r, mat4 m, vec4 c);
+
 #ifdef __cplusplus
 }
 #endif

include/cglm/call/plane.h

@@ -0,0 +1,23 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglmc_plane_h
+#define cglmc_plane_h
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "../cglm.h"
+
+CGLM_EXPORT
+void
+glmc_plane_normalize(vec4 plane);
+
+#ifdef __cplusplus
+}
+#endif
+#endif /* cglmc_plane_h */

include/cglm/call/project.h

@@ -0,0 +1,33 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglmc_project_h
+#define cglmc_project_h
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "../cglm.h"
+
+CGLM_EXPORT
+void
+glmc_unprojecti(vec3 pos, mat4 invMat, vec4 vp, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_unproject(vec3 pos, mat4 m, vec4 vp, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_project(vec3 pos, mat4 m, vec4 vp, vec3 dest);
+
+#ifdef __cplusplus
+}
+#endif
+#endif /* cglmc_project_h */
+
+

include/cglm/call/quat.h

@@ -19,33 +19,83 @@ glmc_quat_identity(versor q);
 
 CGLM_EXPORT
 void
-glmc_quat(versor q,
-          float angle,
-          float x,
-          float y,
-          float z);
+glmc_quat_identity_array(versor * __restrict q, size_t count);
 
 CGLM_EXPORT
 void
-glmc_quatv(versor q,
-          float  angle,
-          vec3   v);
+glmc_quat_init(versor q, float x, float y, float z, float w);
+
+CGLM_EXPORT
+void
+glmc_quat(versor q, float angle, float x, float y, float z);
+
+CGLM_EXPORT
+void
+glmc_quatv(versor q, float angle, vec3 axis);
+
+CGLM_EXPORT
+void
+glmc_quat_copy(versor q, versor dest);
 
 CGLM_EXPORT
 float
 glmc_quat_norm(versor q);
 
+CGLM_EXPORT
+void
+glmc_quat_normalize_to(versor q, versor dest);
+
 CGLM_EXPORT
 void
 glmc_quat_normalize(versor q);
 
 CGLM_EXPORT
 float
-glmc_quat_dot(versor q, versor r);
+glmc_quat_dot(versor p, versor q);
 
 CGLM_EXPORT
 void
-glmc_quat_mulv(versor q1, versor q2, versor dest);
+glmc_quat_conjugate(versor q, versor dest);
+
+CGLM_EXPORT
+void
+glmc_quat_inv(versor q, versor dest);
+
+CGLM_EXPORT
+void
+glmc_quat_add(versor p, versor q, versor dest);
+
+CGLM_EXPORT
+void
+glmc_quat_sub(versor p, versor q, versor dest);
+
+CGLM_EXPORT
+float
+glmc_quat_real(versor q);
+
+CGLM_EXPORT
+void
+glmc_quat_imag(versor q, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_quat_imagn(versor q, vec3 dest);
+
+CGLM_EXPORT
+float
+glmc_quat_imaglen(versor q);
+
+CGLM_EXPORT
+float
+glmc_quat_angle(versor q);
+
+CGLM_EXPORT
+void
+glmc_quat_axis(versor q, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_quat_mul(versor p, versor q, versor dest);
 
 CGLM_EXPORT
 void
@@ -53,10 +103,55 @@ glmc_quat_mat4(versor q, mat4 dest);
 
 CGLM_EXPORT
 void
-glmc_quat_slerp(versor q,
-                versor r,
-                float  t,
-                versor dest);
+glmc_quat_mat4t(versor q, mat4 dest);
+
+CGLM_EXPORT
+void
+glmc_quat_mat3(versor q, mat3 dest);
+
+CGLM_EXPORT
+void
+glmc_quat_mat3t(versor q, mat3 dest);
+
+CGLM_EXPORT
+void
+glmc_quat_lerp(versor from, versor to, float t, versor dest);
+    
+CGLM_EXPORT
+void
+glmc_quat_lerpc(versor from, versor to, float t, versor dest);
+
+CGLM_EXPORT
+void
+glmc_quat_slerp(versor q, versor r, float t, versor dest);
+
+CGLM_EXPORT
+void
+glmc_quat_look(vec3 eye, versor ori, mat4 dest);
+
+CGLM_EXPORT
+void
+glmc_quat_for(vec3 dir, vec3 up, versor dest);
+
+CGLM_EXPORT
+void
+glmc_quat_forp(vec3 from, vec3 to, vec3 up, versor dest);
+
+CGLM_EXPORT
+void
+glmc_quat_rotatev(versor from, vec3 to, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_quat_rotate(mat4 m, versor q, mat4 dest);
+
+CGLM_EXPORT
+void
+glmc_quat_rotate_at(mat4 model, versor q, vec3 pivot);
+
+CGLM_EXPORT
+void
+glmc_quat_rotate_atm(mat4 m, versor q, vec3 pivot);
 
 #ifdef __cplusplus
 }

include/cglm/call/sphere.h

@@ -0,0 +1,39 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglmc_sphere_h
+#define cglmc_sphere_h
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "../cglm.h"
+
+CGLM_EXPORT
+float
+glmc_sphere_radii(vec4 s);
+
+CGLM_EXPORT
+void
+glmc_sphere_transform(vec4 s, mat4 m, vec4 dest);
+
+CGLM_EXPORT
+void
+glmc_sphere_merge(vec4 s1, vec4 s2, vec4 dest);
+
+CGLM_EXPORT
+bool
+glmc_sphere_sphere(vec4 s1, vec4 s2);
+
+CGLM_EXPORT
+bool
+glmc_sphere_point(vec4 s, vec3 point);
+
+#ifdef __cplusplus
+}
+#endif
+#endif /* cglmc_sphere_h */

include/cglm/call/vec2.h

@@ -0,0 +1,155 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglmc_vec2_h
+#define cglmc_vec2_h
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "../cglm.h"
+
+CGLM_EXPORT
+void
+glmc_vec2(float * __restrict v, vec2 dest);
+
+CGLM_EXPORT
+void
+glmc_vec2_copy(vec2 a, vec2 dest);
+
+CGLM_EXPORT
+void
+glmc_vec2_zero(vec2 v);
+
+CGLM_EXPORT
+void
+glmc_vec2_one(vec2 v);
+
+CGLM_EXPORT
+float
+glmc_vec2_dot(vec2 a, vec2 b);
+
+CGLM_EXPORT
+float
+glmc_vec2_cross(vec2 a, vec2 b);
+
+CGLM_EXPORT
+float
+glmc_vec2_norm2(vec2 v);
+
+CGLM_EXPORT
+float
+glmc_vec2_norm(vec2 v);
+
+CGLM_EXPORT
+void
+glmc_vec2_add(vec2 a, vec2 b, vec2 dest);
+
+CGLM_EXPORT
+void
+glmc_vec2_adds(vec2 v, float s, vec2 dest);
+
+CGLM_EXPORT
+void
+glmc_vec2_sub(vec2 a, vec2 b, vec2 dest);
+
+CGLM_EXPORT
+void
+glmc_vec2_subs(vec2 v, float s, vec2 dest);
+
+CGLM_EXPORT
+void
+glmc_vec2_mul(vec2 a, vec2 b, vec2 dest);
+
+CGLM_EXPORT
+void
+glmc_vec2_scale(vec2 v, float s, vec2 dest);
+
+CGLM_EXPORT
+void
+glmc_vec2_scale_as(vec2 v, float s, vec2 dest);
+
+CGLM_EXPORT
+void
+glmc_vec2_div(vec2 a, vec2 b, vec2 dest);
+
+CGLM_EXPORT
+void
+glmc_vec2_divs(vec2 v, float s, vec2 dest);
+
+CGLM_EXPORT
+void
+glmc_vec2_addadd(vec2 a, vec2 b, vec2 dest);
+
+CGLM_EXPORT
+void
+glmc_vec2_subadd(vec2 a, vec2 b, vec2 dest);
+
+CGLM_EXPORT
+void
+glmc_vec2_muladd(vec2 a, vec2 b, vec2 dest);
+
+CGLM_EXPORT
+void
+glmc_vec2_muladds(vec2 a, float s, vec2 dest);
+
+CGLM_EXPORT
+void
+glmc_vec2_maxadd(vec2 a, vec2 b, vec2 dest);
+
+CGLM_EXPORT
+void
+glmc_vec2_minadd(vec2 a, vec2 b, vec2 dest);
+
+CGLM_EXPORT
+void
+glmc_vec2_negate_to(vec2 v, vec2 dest);
+
+CGLM_EXPORT
+void
+glmc_vec2_negate(vec2 v);
+
+CGLM_EXPORT
+void
+glmc_vec2_normalize(vec2 v);
+
+CGLM_EXPORT
+void
+glmc_vec2_normalize_to(vec2 v, vec2 dest);
+
+CGLM_EXPORT
+void
+glmc_vec2_rotate(vec2 v, float angle, vec2 dest);
+
+CGLM_EXPORT
+float
+glmc_vec2_distance2(vec2 a, vec2 b);
+
+CGLM_EXPORT
+float
+glmc_vec2_distance(vec2 a, vec2 b);
+
+CGLM_EXPORT
+void
+glmc_vec2_maxv(vec2 a, vec2 b, vec2 dest);
+
+CGLM_EXPORT
+void
+glmc_vec2_minv(vec2 a, vec2 b, vec2 dest);
+
+CGLM_EXPORT
+void
+glmc_vec2_clamp(vec2 v, float minval, float maxval);
+
+CGLM_EXPORT
+void
+glmc_vec2_lerp(vec2 from, vec2 to, float t, vec2 dest);
+
+#ifdef __cplusplus
+}
+#endif
+#endif /* cglmc_vec2_h */

include/cglm/call/vec3.h

@@ -14,79 +14,297 @@ extern "C" {
 #include "../cglm.h"
 
 /* DEPRECATED! use _copy, _ucopy versions */
-#define glmc_vec_dup(v, dest) glmc_vec_copy(v, dest)
+#define glmc_vec_dup(v, dest)          glmc_vec3_copy(v, dest)
+#define glmc_vec3_flipsign(v)          glmc_vec3_negate(v)
+#define glmc_vec3_flipsign_to(v, dest) glmc_vec3_negate_to(v, dest)
+#define glmc_vec3_inv(v)               glmc_vec3_negate(v)
+#define glmc_vec3_inv_to(v, dest)      glmc_vec3_negate_to(v, dest)
 
 CGLM_EXPORT
 void
-glmc_vec_copy(vec3 a, vec3 dest);
+glmc_vec3(vec4 v4, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_copy(vec3 a, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_zero(vec3 v);
+
+CGLM_EXPORT
+void
+glmc_vec3_one(vec3 v);
 
 CGLM_EXPORT
 float
-glmc_vec_dot(vec3 a, vec3 b);
+glmc_vec3_dot(vec3 a, vec3 b);
 
 CGLM_EXPORT
 void
-glmc_vec_cross(vec3 a, vec3 b, vec3 d);
+glmc_vec3_cross(vec3 a, vec3 b, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_crossn(vec3 a, vec3 b, vec3 dest);
 
 CGLM_EXPORT
 float
-glmc_vec_norm(vec3 vec);
+glmc_vec3_norm(vec3 v);
 
 CGLM_EXPORT
 float
-glmc_vec_norm2(vec3 vec);
-
-CGLM_EXPORT
-void
-glmc_vec_normalize_to(vec3 vec, vec3 dest);
-
-CGLM_EXPORT
-void
-glmc_vec_normalize(vec3 v);
-
-CGLM_EXPORT
-void
-glmc_vec_add(vec3 v1, vec3 v2, vec3 dest);
-
-CGLM_EXPORT
-void
-glmc_vec_sub(vec3 v1, vec3 v2, vec3 dest);
-
-CGLM_EXPORT
-void
-glmc_vec_scale(vec3 v, float s, vec3 dest);
-
-CGLM_EXPORT
-void
-glmc_vec_scale_as(vec3 v, float s, vec3 dest);
-
-CGLM_EXPORT
-void
-glmc_vec_flipsign(vec3 v);
+glmc_vec3_norm2(vec3 v);
     
 CGLM_EXPORT
 float
-glmc_vec_angle(vec3 v1, vec3 v2);
-
-CGLM_EXPORT
-void
-glmc_vec_rotate(vec3 v, float angle, vec3 axis);
-
-CGLM_EXPORT
-void
-glmc_vec_rotate_m4(mat4 m, vec3 v, vec3 dest);
-
-CGLM_EXPORT
-void
-glmc_vec_proj(vec3 a, vec3 b, vec3 dest);
-
-CGLM_EXPORT
-void
-glmc_vec_center(vec3 v1, vec3 v2, vec3 dest);
+glmc_vec3_norm_one(vec3 v);
 
 CGLM_EXPORT
 float
-glmc_vec_distance(vec3 v1, vec3 v2);
+glmc_vec3_norm_inf(vec3 v);
+
+CGLM_EXPORT
+void
+glmc_vec3_normalize_to(vec3 v, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_normalize(vec3 v);
+
+CGLM_EXPORT
+void
+glmc_vec3_add(vec3 a, vec3 b, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_adds(vec3 v, float s, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_sub(vec3 a, vec3 b, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_subs(vec3 v, float s, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_mul(vec3 a, vec3 b, vec3 d);
+
+CGLM_EXPORT
+void
+glmc_vec3_scale(vec3 v, float s, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_scale_as(vec3 v, float s, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_div(vec3 a, vec3 b, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_divs(vec3 a, float s, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_addadd(vec3 a, vec3 b, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_subadd(vec3 a, vec3 b, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_muladd(vec3 a, vec3 b, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_muladds(vec3 a, float s, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_maxadd(vec3 a, vec3 b, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_minadd(vec3 a, vec3 b, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_negate(vec3 v);
+
+CGLM_EXPORT
+void
+glmc_vec3_negate_to(vec3 v, vec3 dest);
+
+CGLM_EXPORT
+float
+glmc_vec3_angle(vec3 a, vec3 b);
+
+CGLM_EXPORT
+void
+glmc_vec3_rotate(vec3 v, float angle, vec3 axis);
+
+CGLM_EXPORT
+void
+glmc_vec3_rotate_m4(mat4 m, vec3 v, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_rotate_m3(mat3 m, vec3 v, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_proj(vec3 a, vec3 b, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_center(vec3 a, vec3 b, vec3 dest);
+
+CGLM_EXPORT
+float
+glmc_vec3_distance2(vec3 a, vec3 b);
+
+CGLM_EXPORT
+float
+glmc_vec3_distance(vec3 a, vec3 b);
+
+CGLM_EXPORT
+void
+glmc_vec3_maxv(vec3 a, vec3 b, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_minv(vec3 a, vec3 b, vec3 dest);
+
+CGLM_EXPORT
+void
+glmc_vec3_clamp(vec3 v, float minVal, float maxVal);
+
+CGLM_EXPORT
+void
+glmc_vec3_ortho(vec3 v, vec3 dest);
+    
+CGLM_EXPORT
+void
+glmc_vec3_lerp(vec3 from, vec3 to, float t, vec3 dest);
+    
+CGLM_EXPORT
+void
+glmc_vec3_lerpc(vec3 from, vec3 to, float t, vec3 dest);
+    
+CGLM_INLINE
+void
+glmc_vec3_mix(vec3 from, vec3 to, float t, vec3 dest) {
+  glmc_vec3_lerp(from, to, t, dest);
+}
+
+CGLM_INLINE
+void
+glmc_vec3_mixc(vec3 from, vec3 to, float t, vec3 dest) {
+  glmc_vec3_lerpc(from, to, t, dest);
+}
+    
+CGLM_EXPORT
+void
+glmc_vec3_step_uni(float edge, vec3 x, vec3 dest);
+    
+CGLM_EXPORT
+void
+glmc_vec3_step(vec3 edge, vec3 x, vec3 dest);
+    
+CGLM_EXPORT
+void
+glmc_vec3_smoothstep_uni(float edge0, float edge1, vec3 x, vec3 dest);
+    
+CGLM_EXPORT
+void
+glmc_vec3_smoothstep(vec3 edge0, vec3 edge1, vec3 x, vec3 dest);
+    
+CGLM_EXPORT
+void
+glmc_vec3_smoothinterp(vec3 from, vec3 to, float t, vec3 dest);
+    
+CGLM_EXPORT
+void
+glmc_vec3_smoothinterpc(vec3 from, vec3 to, float t, vec3 dest);
+
+/* ext */
+
+CGLM_EXPORT
+void
+glmc_vec3_mulv(vec3 a, vec3 b, vec3 d);
+
+CGLM_EXPORT
+void
+glmc_vec3_broadcast(float val, vec3 d);
+    
+CGLM_EXPORT
+void
+glmc_vec3_fill(vec3 v, float val);
+
+CGLM_EXPORT
+bool
+glmc_vec3_eq(vec3 v, float val);
+
+CGLM_EXPORT
+bool
+glmc_vec3_eq_eps(vec3 v, float val);
+
+CGLM_EXPORT
+bool
+glmc_vec3_eq_all(vec3 v);
+
+CGLM_EXPORT
+bool
+glmc_vec3_eqv(vec3 a, vec3 b);
+
+CGLM_EXPORT
+bool
+glmc_vec3_eqv_eps(vec3 a, vec3 b);
+
+CGLM_EXPORT
+float
+glmc_vec3_max(vec3 v);
+
+CGLM_EXPORT
+float
+glmc_vec3_min(vec3 v);
+
+CGLM_EXPORT
+bool
+glmc_vec3_isnan(vec3 v);
+
+CGLM_EXPORT
+bool
+glmc_vec3_isinf(vec3 v);
+
+CGLM_EXPORT
+bool
+glmc_vec3_isvalid(vec3 v);
+
+CGLM_EXPORT
+void
+glmc_vec3_sign(vec3 v, vec3 dest);
+    
+CGLM_EXPORT
+void
+glmc_vec3_abs(vec3 v, vec3 dest);
+    
+CGLM_EXPORT
+void
+glmc_vec3_fract(vec3 v, vec3 dest);
+    
+CGLM_EXPORT
+float
+glmc_vec3_hadd(vec3 v);
+
+CGLM_EXPORT
+void
+glmc_vec3_sqrt(vec3 v, vec3 dest);
 
 #ifdef __cplusplus
 }

include/cglm/call/vec4.h

@@ -14,32 +14,60 @@ extern "C" {
 #include "../cglm.h"
 
 /* DEPRECATED! use _copy, _ucopy versions */
-#define glmc_vec4_dup3(v, dest) glmc_vec4_copy3(v, dest)
-#define glmc_vec4_dup(v, dest)  glmc_vec4_copy(v, dest)
+#define glmc_vec4_dup3(v, dest)         glmc_vec4_copy3(v, dest)
+#define glmc_vec4_dup(v, dest)          glmc_vec4_copy(v, dest)
+#define glmc_vec4_flipsign(v)           glmc_vec4_negate(v)
+#define glmc_vec4_flipsign_to(v, dest)  glmc_vec4_negate_to(v, dest)
+#define glmc_vec4_inv(v)                glmc_vec4_negate(v)
+#define glmc_vec4_inv_to(v, dest)       glmc_vec4_negate_to(v, dest)
 
 CGLM_EXPORT
 void
-glmc_vec4_copy3(vec4 a, vec3 dest);
+glmc_vec4(vec3 v3, float last, vec4 dest);
+
+CGLM_EXPORT
+void
+glmc_vec4_zero(vec4 v);
+
+CGLM_EXPORT
+void
+glmc_vec4_one(vec4 v);
+
+CGLM_EXPORT
+void
+glmc_vec4_copy3(vec4 v, vec3 dest);
 
 CGLM_EXPORT
 void
 glmc_vec4_copy(vec4 v, vec4 dest);
 
+CGLM_EXPORT
+void
+glmc_vec4_ucopy(vec4 v, vec4 dest);
+
 CGLM_EXPORT
 float
 glmc_vec4_dot(vec4 a, vec4 b);
 
 CGLM_EXPORT
 float
-glmc_vec4_norm(vec4 vec);
+glmc_vec4_norm(vec4 v);
 
 CGLM_EXPORT
 float
-glmc_vec4_norm2(vec4 vec);
+glmc_vec4_norm2(vec4 v);
+    
+CGLM_EXPORT
+float
+glmc_vec4_norm_one(vec4 v);
+
+CGLM_EXPORT
+float
+glmc_vec4_norm_inf(vec4 v);
 
 CGLM_EXPORT
 void
-glmc_vec4_normalize_to(vec4 vec, vec4 dest);
+glmc_vec4_normalize_to(vec4 v, vec4 dest);
 
 CGLM_EXPORT
 void
@@ -47,11 +75,23 @@ glmc_vec4_normalize(vec4 v);
 
 CGLM_EXPORT
 void
-glmc_vec4_add(vec4 v1, vec4 v2, vec4 dest);
+glmc_vec4_add(vec4 a, vec4 b, vec4 dest);
 
 CGLM_EXPORT
 void
-glmc_vec4_sub(vec4 v1, vec4 v2, vec4 dest);
+glmc_vec4_adds(vec4 v, float s, vec4 dest);
+
+CGLM_EXPORT
+void
+glmc_vec4_sub(vec4 a, vec4 b, vec4 dest);
+
+CGLM_EXPORT
+void
+glmc_vec4_subs(vec4 v, float s, vec4 dest);
+
+CGLM_EXPORT
+void
+glmc_vec4_mul(vec4 a, vec4 b, vec4 d);
 
 CGLM_EXPORT
 void
@@ -63,11 +103,185 @@ glmc_vec4_scale_as(vec3 v, float s, vec3 dest);
 
 CGLM_EXPORT
 void
-glmc_vec4_flipsign(vec4 v);
+glmc_vec4_div(vec4 a, vec4 b, vec4 dest);
+
+CGLM_EXPORT
+void
+glmc_vec4_divs(vec4 v, float s, vec4 dest);
+
+CGLM_EXPORT
+void
+glmc_vec4_addadd(vec4 a, vec4 b, vec4 dest);
+
+CGLM_EXPORT
+void
+glmc_vec4_subadd(vec4 a, vec4 b, vec4 dest);
+
+CGLM_EXPORT
+void
+glmc_vec4_muladd(vec4 a, vec4 b, vec4 dest);
+
+CGLM_EXPORT
+void
+glmc_vec4_muladds(vec4 a, float s, vec4 dest);
+
+CGLM_EXPORT
+void
+glmc_vec4_maxadd(vec4 a, vec4 b, vec4 dest);
+
+CGLM_EXPORT
+void
+glmc_vec4_minadd(vec4 a, vec4 b, vec4 dest);
+
+CGLM_EXPORT
+void
+glmc_vec4_negate(vec4 v);
+
+CGLM_EXPORT
+void
+glmc_vec4_negate_to(vec4 v, vec4 dest);
     
 CGLM_EXPORT
 float
-glmc_vec4_distance(vec4 v1, vec4 v2);
+glmc_vec4_distance(vec4 a, vec4 b);
+    
+CGLM_EXPORT
+float
+glmc_vec4_distance2(vec4 a, vec4 b);
+
+CGLM_EXPORT
+void
+glmc_vec4_maxv(vec4 a, vec4 b, vec4 dest);
+
+CGLM_EXPORT
+void
+glmc_vec4_minv(vec4 a, vec4 b, vec4 dest);
+
+CGLM_EXPORT
+void
+glmc_vec4_clamp(vec4 v, float minVal, float maxVal);
+    
+CGLM_EXPORT
+void
+glmc_vec4_lerp(vec4 from, vec4 to, float t, vec4 dest);
+    
+CGLM_EXPORT
+void
+glmc_vec4_lerpc(vec4 from, vec4 to, float t, vec4 dest);
+    
+CGLM_INLINE
+void
+glmc_vec4_mix(vec4 from, vec4 to, float t, vec4 dest) {
+  glmc_vec4_lerp(from, to, t, dest);
+}
+
+CGLM_INLINE
+void
+glmc_vec4_mixc(vec4 from, vec4 to, float t, vec4 dest) {
+  glmc_vec4_lerpc(from, to, t, dest);
+}
+    
+CGLM_EXPORT
+void
+glmc_vec4_step_uni(float edge, vec4 x, vec4 dest);
+    
+CGLM_EXPORT
+void
+glmc_vec4_step(vec4 edge, vec4 x, vec4 dest);
+    
+CGLM_EXPORT
+void
+glmc_vec4_smoothstep_uni(float edge0, float edge1, vec4 x, vec4 dest);
+    
+CGLM_EXPORT
+void
+glmc_vec4_smoothstep(vec4 edge0, vec4 edge1, vec4 x, vec4 dest);
+    
+CGLM_EXPORT
+void
+glmc_vec4_smoothinterp(vec4 from, vec4 to, float t, vec4 dest);
+    
+CGLM_EXPORT
+void
+glmc_vec4_smoothinterpc(vec4 from, vec4 to, float t, vec4 dest);
+
+CGLM_EXPORT
+void
+glmc_vec4_cubic(float s, vec4 dest);
+
+/* ext */
+
+CGLM_EXPORT
+void
+glmc_vec4_mulv(vec4 a, vec4 b, vec4 d);
+
+CGLM_EXPORT
+void
+glmc_vec4_broadcast(float val, vec4 d);
+    
+CGLM_EXPORT
+void
+glmc_vec4_fill(vec4 v, float val);
+
+CGLM_EXPORT
+bool
+glmc_vec4_eq(vec4 v, float val);
+
+CGLM_EXPORT
+bool
+glmc_vec4_eq_eps(vec4 v, float val);
+
+CGLM_EXPORT
+bool
+glmc_vec4_eq_all(vec4 v);
+
+CGLM_EXPORT
+bool
+glmc_vec4_eqv(vec4 a, vec4 b);
+
+CGLM_EXPORT
+bool
+glmc_vec4_eqv_eps(vec4 a, vec4 b);
+
+CGLM_EXPORT
+float
+glmc_vec4_max(vec4 v);
+
+CGLM_EXPORT
+float
+glmc_vec4_min(vec4 v);
+
+CGLM_EXPORT
+bool
+glmc_vec4_isnan(vec4 v);
+
+CGLM_EXPORT
+bool
+glmc_vec4_isinf(vec4 v);
+
+CGLM_EXPORT
+bool
+glmc_vec4_isvalid(vec4 v);
+
+CGLM_EXPORT
+void
+glmc_vec4_sign(vec4 v, vec4 dest);
+    
+CGLM_EXPORT
+void
+glmc_vec4_abs(vec4 v, vec4 dest);
+    
+CGLM_EXPORT
+void
+glmc_vec4_fract(vec4 v, vec4 dest);
+    
+CGLM_EXPORT
+float
+glmc_vec4_hadd(vec4 v);
+
+CGLM_EXPORT
+void
+glmc_vec4_sqrt(vec4 v, vec4 dest);
 
 #ifdef __cplusplus
 }

include/cglm/cam.h

@@ -7,36 +7,49 @@
 
 /*
  Functions:
-   CGLM_INLINE void glm_frustum(float left,
-                                float right,
-                                float bottom,
-                                float top,
-                                float near,
-                                float far,
-                                mat4  dest);
-   CGLM_INLINE void glm_ortho(float left,
-                              float right,
-                              float bottom,
-                              float top,
-                              float near,
-                              float far,
-                              mat4  dest);
-   CGLM_INLINE void glm_ortho_default(float aspect, mat4  dest);
-   CGLM_INLINE void glm_ortho_default_s(float aspect, float size, mat4  dest);
-   CGLM_INLINE void glm_perspective(float fovy,
-                                    float aspect,
-                                    float near,
-                                    float far,
-                                    mat4  dest);
-   CGLM_INLINE void glm_perspective_default(float aspect, mat4  dest);
-   CGLM_INLINE void glm_perspective_resize(float aspect, mat4  proj);
-   CGLM_INLINE void glm_lookat(vec3 eye, vec3 center, vec3 up, mat4 dest);
+   CGLM_INLINE void  glm_frustum(float left,    float right,
+                                 float bottom,  float top,
+                                 float nearVal, float farVal,
+                                 mat4  dest)
+   CGLM_INLINE void  glm_ortho(float left,    float right,
+                               float bottom,  float top,
+                               float nearVal, float farVal,
+                               mat4  dest)
+   CGLM_INLINE void  glm_ortho_aabb(vec3 box[2], mat4 dest)
+   CGLM_INLINE void  glm_ortho_aabb_p(vec3 box[2],  float padding, mat4 dest)
+   CGLM_INLINE void  glm_ortho_aabb_pz(vec3 box[2], float padding, mat4 dest)
+   CGLM_INLINE void  glm_ortho_default(float aspect, mat4  dest)
+   CGLM_INLINE void  glm_ortho_default_s(float aspect, float size, mat4 dest)
+   CGLM_INLINE void  glm_perspective(float fovy,
+                                     float aspect,
+                                     float nearVal,
+                                     float farVal,
+                                     mat4  dest)
+   CGLM_INLINE void  glm_perspective_default(float aspect, mat4 dest)
+   CGLM_INLINE void  glm_perspective_resize(float aspect, mat4 proj)
+   CGLM_INLINE void  glm_lookat(vec3 eye, vec3 center, vec3 up, mat4 dest)
+   CGLM_INLINE void  glm_look(vec3 eye, vec3 dir, vec3 up, mat4 dest)
+   CGLM_INLINE void  glm_look_anyup(vec3 eye, vec3 dir, mat4 dest)
+   CGLM_INLINE void  glm_persp_decomp(mat4   proj,
+                                      float *nearVal, float *farVal,
+                                      float *top,     float *bottom,
+                                      float *left,    float *right)
+   CGLM_INLINE void  glm_persp_decompv(mat4 proj, float dest[6])
+   CGLM_INLINE void  glm_persp_decomp_x(mat4 proj, float *left, float *right)
+   CGLM_INLINE void  glm_persp_decomp_y(mat4 proj, float *top,  float *bottom)
+   CGLM_INLINE void  glm_persp_decomp_z(mat4 proj, float *nearv, float *farv)
+   CGLM_INLINE void  glm_persp_decomp_far(mat4 proj, float *farVal)
+   CGLM_INLINE void  glm_persp_decomp_near(mat4 proj, float *nearVal)
+   CGLM_INLINE float glm_persp_fovy(mat4 proj)
+   CGLM_INLINE float glm_persp_aspect(mat4 proj)
+   CGLM_INLINE void  glm_persp_sizes(mat4 proj, float fovy, vec4 dest)
  */
 
 #ifndef cglm_vcam_h
 #define cglm_vcam_h
 
 #include "common.h"
+#include "plane.h"
 
 /*!
  * @brief set up perspective peprojection matrix
@@ -51,16 +64,13 @@
  */
 CGLM_INLINE
 void
-glm_frustum(float left,
-            float right,
-            float bottom,
-            float top,
-            float nearVal,
-            float farVal,
+glm_frustum(float left,    float right,
+            float bottom,  float top,
+            float nearVal, float farVal,
             mat4  dest) {
   float rl, tb, fn, nv;
 
-  glm__memzero(float, dest, sizeof(mat4));
+  glm_mat4_zero(dest);
 
   rl = 1.0f / (right  - left);
   tb = 1.0f / (top    - bottom);
@@ -89,16 +99,13 @@ glm_frustum(float left,
  */
 CGLM_INLINE
 void
-glm_ortho(float left,
-          float right,
-          float bottom,
-          float top,
-          float nearVal,
-          float farVal,
+glm_ortho(float left,    float right,
+          float bottom,  float top,
+          float nearVal, float farVal,
           mat4  dest) {
   float rl, tb, fn;
 
-  glm__memzero(float, dest, sizeof(mat4));
+  glm_mat4_zero(dest);
 
   rl = 1.0f / (right  - left);
   tb = 1.0f / (top    - bottom);
@@ -113,6 +120,59 @@ glm_ortho(float left,
   dest[3][3] = 1.0f;
 }
 
+/*!
+ * @brief set up orthographic projection matrix using bounding box
+ *
+ * bounding box (AABB) must be in view space
+ *
+ * @param[in]  box   AABB
+ * @param[out] dest  result matrix
+ */
+CGLM_INLINE
+void
+glm_ortho_aabb(vec3 box[2], mat4 dest) {
+  glm_ortho(box[0][0],  box[1][0],
+            box[0][1],  box[1][1],
+           -box[1][2], -box[0][2],
+            dest);
+}
+
+/*!
+ * @brief set up orthographic projection matrix using bounding box
+ *
+ * bounding box (AABB) must be in view space
+ *
+ * @param[in]  box     AABB
+ * @param[in]  padding padding
+ * @param[out] dest    result matrix
+ */
+CGLM_INLINE
+void
+glm_ortho_aabb_p(vec3 box[2], float padding, mat4 dest) {
+  glm_ortho(box[0][0] - padding,    box[1][0] + padding,
+            box[0][1] - padding,    box[1][1] + padding,
+          -(box[1][2] + padding), -(box[0][2] - padding),
+            dest);
+}
+
+/*!
+ * @brief set up orthographic projection matrix using bounding box
+ *
+ * bounding box (AABB) must be in view space
+ *
+ * @param[in]  box     AABB
+ * @param[in]  padding padding for near and far
+ * @param[out] dest    result matrix
+ */
+CGLM_INLINE
+void
+glm_ortho_aabb_pz(vec3 box[2], float padding, mat4 dest) {
+  glm_ortho(box[0][0],              box[1][0],
+            box[0][1],              box[1][1],
+          -(box[1][2] + padding), -(box[0][2] - padding),
+            dest);
+}
+
 /*!
  * @brief set up unit orthographic projection matrix
  *
@@ -121,26 +181,15 @@ glm_ortho(float left,
  */
 CGLM_INLINE
 void
-glm_ortho_default(float aspect,
-                  mat4  dest) {
+glm_ortho_default(float aspect, mat4 dest) {
   if (aspect >= 1.0f) {
-    glm_ortho(-1.0f * aspect,
-               1.0f * aspect,
-              -1.0f,
-               1.0f,
-              -100.0f,
-               100.0f,
-               dest);
-	  return;
+    glm_ortho(-aspect, aspect, -1.0f, 1.0f, -100.0f, 100.0f, dest);
+    return;
   }
 
-  glm_ortho(-1.0f,
-             1.0f,
-            -1.0f / aspect,
-             1.0f / aspect,
-            -100.0f,
-             100.0f,
-             dest);
+  aspect = 1.0f / aspect;
+
+  glm_ortho(-1.0f, 1.0f, -aspect, aspect, -100.0f, 100.0f, dest);
 }
 
 /*!
@@ -152,9 +201,7 @@ glm_ortho_default(float aspect,
  */
 CGLM_INLINE
 void
-glm_ortho_default_s(float aspect,
-                    float size,
-                    mat4  dest) {
+glm_ortho_default_s(float aspect, float size, mat4 dest) {
   if (aspect >= 1.0f) {
     glm_ortho(-size * aspect,
                size * aspect,
@@ -163,7 +210,7 @@ glm_ortho_default_s(float aspect,
               -size - 100.0f,
                size + 100.0f,
                dest);
-	  return;
+    return;
   }
 
   glm_ortho(-size,
@@ -193,7 +240,7 @@ glm_perspective(float fovy,
                 mat4  dest) {
   float f, fn;
 
-  glm__memzero(float, dest, sizeof(mat4));
+  glm_mat4_zero(dest);
 
   f  = 1.0f / tanf(fovy * 0.5f);
   fn = 1.0f / (nearVal - farVal);
@@ -205,6 +252,30 @@ glm_perspective(float fovy,
   dest[3][2] = 2.0f * nearVal * farVal * fn;
 }
 
+/*!
+ * @brief extend perspective projection matrix's far distance
+ *
+ * this function does not guarantee far >= near, be aware of that!
+ *
+ * @param[in, out] proj      projection matrix to extend
+ * @param[in]      deltaFar  distance from existing far (negative to shink)
+ */
+CGLM_INLINE
+void
+glm_persp_move_far(mat4 proj, float deltaFar) {
+  float fn, farVal, nearVal, p22, p32;
+
+  p22        = proj[2][2];
+  p32        = proj[3][2];
+
+  nearVal    = p32 / (p22 - 1.0f);
+  farVal     = p32 / (p22 + 1.0f) + deltaFar;
+  fn         = 1.0f / (nearVal - farVal);
+
+  proj[2][2] = (nearVal + farVal) * fn;
+  proj[3][2] = 2.0f * nearVal * farVal * fn;
+}
+
 /*!
  * @brief set up perspective projection matrix with default near/far
  *        and angle values
@@ -214,18 +285,13 @@ glm_perspective(float fovy,
  */
 CGLM_INLINE
 void
-glm_perspective_default(float aspect,
-                        mat4  dest) {
-  glm_perspective((float)CGLM_PI_4,
-                  aspect,
-                  0.01f,
-                  100.0f,
-                  dest);
+glm_perspective_default(float aspect, mat4 dest) {
+  glm_perspective(GLM_PI_4f, aspect, 0.01f, 100.0f, dest);
 }
 
 /*!
  * @brief resize perspective matrix by aspect ratio ( width / height )
- *        this very make easy to resize proj matrix when window, viewport
+ *        this makes very easy to resize proj matrix when window /viewport
  *        reized
  *
  * @param[in]      aspect aspect ratio ( width / height )
@@ -233,8 +299,7 @@ glm_perspective_default(float aspect,
  */
 CGLM_INLINE
 void
-glm_perspective_resize(float aspect,
-                       mat4  proj) {
+glm_perspective_resize(float aspect, mat4 proj) {
   if (proj[0][0] == 0.0f)
     return;
 
@@ -244,6 +309,9 @@ glm_perspective_resize(float aspect,
 /*!
  * @brief set up view matrix
  *
+ * NOTE: The UP vector must not be parallel to the line of sight from
+ *       the eye point to the reference point
+ *
  * @param[in]  eye    eye vector
  * @param[in]  center center vector
  * @param[in]  up     up vector
@@ -251,19 +319,14 @@ glm_perspective_resize(float aspect,
  */
 CGLM_INLINE
 void
-glm_lookat(vec3 eye,
-           vec3 center,
-           vec3 up,
-           mat4 dest) {
-  vec3 f, u, s;
+glm_lookat(vec3 eye, vec3 center, vec3 up, mat4 dest) {
+  CGLM_ALIGN(8) vec3 f, u, s;
 
-  glm_vec_sub(center, eye, f);
-  glm_vec_normalize(f);
+  glm_vec3_sub(center, eye, f);
+  glm_vec3_normalize(f);
 
-  glm_vec_cross(f, up, s);
-  glm_vec_normalize(s);
-
-  glm_vec_cross(s, f, u);
+  glm_vec3_crossn(f, up, s);
+  glm_vec3_cross(s, f, u);
 
   dest[0][0] = s[0];
   dest[0][1] = u[0];
@@ -274,11 +337,246 @@ glm_lookat(vec3 eye,
   dest[2][0] = s[2];
   dest[2][1] = u[2];
   dest[2][2] =-f[2];
-  dest[3][0] =-glm_vec_dot(s, eye);
-  dest[3][1] =-glm_vec_dot(u, eye);
-  dest[3][2] = glm_vec_dot(f, eye);
+  dest[3][0] =-glm_vec3_dot(s, eye);
+  dest[3][1] =-glm_vec3_dot(u, eye);
+  dest[3][2] = glm_vec3_dot(f, eye);
   dest[0][3] = dest[1][3] = dest[2][3] = 0.0f;
   dest[3][3] = 1.0f;
 }
 
+/*!
+ * @brief set up view matrix
+ *
+ * convenient wrapper for lookat: if you only have direction not target self
+ * then this might be useful. Because you need to get target from direction.
+ *
+ * NOTE: The UP vector must not be parallel to the line of sight from
+ *       the eye point to the reference point
+ *
+ * @param[in]  eye    eye vector
+ * @param[in]  dir    direction vector
+ * @param[in]  up     up vector
+ * @param[out] dest   result matrix
+ */
+CGLM_INLINE
+void
+glm_look(vec3 eye, vec3 dir, vec3 up, mat4 dest) {
+  CGLM_ALIGN(8) vec3 target;
+  glm_vec3_add(eye, dir, target);
+  glm_lookat(eye, target, up, dest);
+}
+
+/*!
+ * @brief set up view matrix
+ *
+ * convenient wrapper for look: if you only have direction and if you don't
+ * care what UP vector is then this might be useful to create view matrix
+ *
+ * @param[in]  eye    eye vector
+ * @param[in]  dir    direction vector
+ * @param[out] dest   result matrix
+ */
+CGLM_INLINE
+void
+glm_look_anyup(vec3 eye, vec3 dir, mat4 dest) {
+  CGLM_ALIGN(8) vec3 up;
+  glm_vec3_ortho(dir, up);
+  glm_look(eye, dir, up, dest);
+}
+
+/*!
+ * @brief decomposes frustum values of perspective projection.
+ *
+ * @param[in]  proj    perspective projection matrix
+ * @param[out] nearVal near
+ * @param[out] farVal  far
+ * @param[out] top     top
+ * @param[out] bottom  bottom
+ * @param[out] left    left
+ * @param[out] right   right
+ */
+CGLM_INLINE
+void
+glm_persp_decomp(mat4 proj,
+                 float * __restrict nearVal, float * __restrict farVal,
+                 float * __restrict top,     float * __restrict bottom,
+                 float * __restrict left,    float * __restrict right) {
+  float m00, m11, m20, m21, m22, m32, n, f;
+  float n_m11, n_m00;
+
+  m00 = proj[0][0];
+  m11 = proj[1][1];
+  m20 = proj[2][0];
+  m21 = proj[2][1];
+  m22 = proj[2][2];
+  m32 = proj[3][2];
+
+  n = m32 / (m22 - 1.0f);
+  f = m32 / (m22 + 1.0f);
+
+  n_m11 = n / m11;
+  n_m00 = n / m00;
+
+  *nearVal = n;
+  *farVal  = f;
+  *bottom  = n_m11 * (m21 - 1.0f);
+  *top     = n_m11 * (m21 + 1.0f);
+  *left    = n_m00 * (m20 - 1.0f);
+  *right   = n_m00 * (m20 + 1.0f);
+}
+
+/*!
+ * @brief decomposes frustum values of perspective projection.
+ *        this makes easy to get all values at once
+ *
+ * @param[in]  proj   perspective projection matrix
+ * @param[out] dest   array
+ */
+CGLM_INLINE
+void
+glm_persp_decompv(mat4 proj, float dest[6]) {
+  glm_persp_decomp(proj, &dest[0], &dest[1], &dest[2],
+                         &dest[3], &dest[4], &dest[5]);
+}
+
+/*!
+ * @brief decomposes left and right values of perspective projection.
+ *        x stands for x axis (left / right axis)
+ *
+ * @param[in]  proj  perspective projection matrix
+ * @param[out] left  left
+ * @param[out] right right
+ */
+CGLM_INLINE
+void
+glm_persp_decomp_x(mat4 proj,
+                   float * __restrict left,
+                   float * __restrict right) {
+  float nearVal, m20, m00;
+
+  m00 = proj[0][0];
+  m20 = proj[2][0];
+
+  nearVal = proj[3][2] / (proj[3][3] - 1.0f);
+  *left   = nearVal * (m20 - 1.0f) / m00;
+  *right  = nearVal * (m20 + 1.0f) / m00;
+}
+
+/*!
+ * @brief decomposes top and bottom values of perspective projection.
+ *        y stands for y axis (top / botom axis)
+ *
+ * @param[in]  proj   perspective projection matrix
+ * @param[out] top    top
+ * @param[out] bottom bottom
+ */
+CGLM_INLINE
+void
+glm_persp_decomp_y(mat4 proj,
+                   float * __restrict top,
+                   float * __restrict bottom) {
+  float nearVal, m21, m11;
+
+  m21 = proj[2][1];
+  m11 = proj[1][1];
+
+  nearVal = proj[3][2] / (proj[3][3] - 1.0f);
+  *bottom = nearVal * (m21 - 1) / m11;
+  *top    = nearVal * (m21 + 1) / m11;
+}
+
+/*!
+ * @brief decomposes near and far values of perspective projection.
+ *        z stands for z axis (near / far axis)
+ *
+ * @param[in]  proj    perspective projection matrix
+ * @param[out] nearVal near
+ * @param[out] farVal  far
+ */
+CGLM_INLINE
+void
+glm_persp_decomp_z(mat4 proj,
+                   float * __restrict nearVal,
+                   float * __restrict farVal) {
+  float m32, m22;
+
+  m32 = proj[3][2];
+  m22 = proj[2][2];
+
+  *nearVal = m32 / (m22 - 1.0f);
+  *farVal  = m32 / (m22 + 1.0f);
+}
+
+/*!
+ * @brief decomposes far value of perspective projection.
+ *
+ * @param[in]  proj   perspective projection matrix
+ * @param[out] farVal far
+ */
+CGLM_INLINE
+void
+glm_persp_decomp_far(mat4 proj, float * __restrict farVal) {
+  *farVal = proj[3][2] / (proj[2][2] + 1.0f);
+}
+
+/*!
+ * @brief decomposes near value of perspective projection.
+ *
+ * @param[in]  proj    perspective projection matrix
+ * @param[out] nearVal near
+ */
+CGLM_INLINE
+void
+glm_persp_decomp_near(mat4 proj, float * __restrict nearVal) {
+  *nearVal = proj[3][2] / (proj[2][2] - 1.0f);
+}
+
+/*!
+ * @brief returns field of view angle along the Y-axis (in radians)
+ *
+ * if you need to degrees, use glm_deg to convert it or use this:
+ * fovy_deg = glm_deg(glm_persp_fovy(projMatrix))
+ *
+ * @param[in] proj perspective projection matrix
+ */
+CGLM_INLINE
+float
+glm_persp_fovy(mat4 proj) {
+  return 2.0f * atanf(1.0f / proj[1][1]);
+}
+
+/*!
+ * @brief returns aspect ratio of perspective projection
+ *
+ * @param[in] proj perspective projection matrix
+ */
+CGLM_INLINE
+float
+glm_persp_aspect(mat4 proj) {
+  return proj[1][1] / proj[0][0];
+}
+
+/*!
+ * @brief returns sizes of near and far planes of perspective projection
+ *
+ * @param[in]  proj perspective projection matrix
+ * @param[in]  fovy fovy (see brief)
+ * @param[out] dest sizes order: [Wnear, Hnear, Wfar, Hfar]
+ */
+CGLM_INLINE
+void
+glm_persp_sizes(mat4 proj, float fovy, vec4 dest) {
+  float t, a, nearVal, farVal;
+
+  t = 2.0f * tanf(fovy * 0.5f);
+  a = glm_persp_aspect(proj);
+
+  glm_persp_decomp_z(proj, &nearVal, &farVal);
+
+  dest[1]  = t * nearVal;
+  dest[3]  = t * farVal;
+  dest[0]  = a * dest[1];
+  dest[2]  = a * dest[3];
+}
+
 #endif /* cglm_vcam_h */

include/cglm/cglm.h

@@ -9,15 +9,26 @@
 #define cglm_h
 
 #include "common.h"
+#include "vec2.h"
 #include "vec3.h"
 #include "vec4.h"
 #include "mat4.h"
 #include "mat3.h"
+#include "mat2.h"
 #include "affine.h"
 #include "cam.h"
+#include "frustum.h"
 #include "quat.h"
 #include "euler.h"
+#include "plane.h"
+#include "box.h"
+#include "color.h"
 #include "util.h"
 #include "io.h"
+#include "project.h"
+#include "sphere.h"
+#include "ease.h"
+#include "curve.h"
+#include "bezier.h"
 
 #endif /* cglm_h */

include/cglm/color.h

@@ -0,0 +1,26 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglm_color_h
+#define cglm_color_h
+
+#include "common.h"
+#include "vec3.h"
+
+/*!
+ * @brief averages the color channels into one value
+ *
+ * @param[in]  rgb RGB color
+ */
+CGLM_INLINE
+float
+glm_luminance(vec3 rgb) {
+  vec3 l = {0.212671f, 0.715160f, 0.072169f};
+  return glm_dot(rgb, l);
+}
+
+#endif /* cglm_color_h */

include/cglm/common.h

@@ -11,9 +11,12 @@
 #define _USE_MATH_DEFINES /* for windows */
 
 #include <stdint.h>
+#include <stddef.h>
 #include <math.h>
+#include <float.h>
+#include <stdbool.h>
 
-#if defined(_WIN32)
+#if defined(_MSC_VER)
 #  ifdef CGLM_DLL
 #    define CGLM_EXPORT __declspec(dllexport)
 #  else
@@ -25,33 +28,8 @@
 #  define CGLM_INLINE static inline __attribute((always_inline))
 #endif
 
-#define glm__memcpy(type, dest, src, size)                                    \
-  do {                                                                        \
-    type *srci;                                                               \
-    type *srci_end;                                                           \
-    type *desti;                                                              \
-                                                                              \
-    srci     = (type *)src;                                                   \
-    srci_end = (type *)((char *)srci + size);                                 \
-    desti    = (type *)dest;                                                  \
-                                                                              \
-    while (srci != srci_end)                                                  \
-      *desti++ = *srci++;                                                     \
-  } while (0)
-
-#define glm__memset(type, dest, size, val)                                    \
-  do {                                                                        \
-    type *desti;                                                              \
-    type *desti_end;                                                          \
-                                                                              \
-    desti     = (type *)dest;                                                 \
-    desti_end = (type *)((char *)desti + size);                               \
-                                                                              \
-    while (desti != desti_end)                                                \
-      *desti++ = val;                                                         \
-  } while (0)
-
-#define glm__memzero(type, dest, size) glm__memset(type, dest, size, 0)
+#define GLM_SHUFFLE4(z, y, x, w) (((z) << 6) | ((y) << 4) | ((x) << 2) | (w))
+#define GLM_SHUFFLE3(z, y, x)    (((z) << 4) | ((y) << 2) | (x))
 
 #include "types.h"
 #include "simd/intrin.h"

include/cglm/curve.h

@@ -0,0 +1,40 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglm_curve_h
+#define cglm_curve_h
+
+#include "common.h"
+#include "vec4.h"
+#include "mat4.h"
+
+/*!
+ * @brief helper function to calculate S*M*C multiplication for curves
+ *
+ * This function does not encourage you to use SMC,
+ * instead it is a helper if you use SMC.
+ *
+ * if you want to specify S as vector then use more generic glm_mat4_rmc() func.
+ *
+ * Example usage:
+ *  B(s) = glm_smc(s, GLM_BEZIER_MAT, (vec4){p0, c0, c1, p1})
+ *
+ * @param[in]  s  parameter between 0 and 1 (this will be [s3, s2, s, 1])
+ * @param[in]  m  basis matrix
+ * @param[in]  c  position/control vector
+ *
+ * @return B(s)
+ */
+CGLM_INLINE
+float
+glm_smc(float s, mat4 m, vec4 c) {
+  vec4 vs;
+  glm_vec4_cubic(s, vs);
+  return glm_mat4_rmc(vs, m, c);
+}
+
+#endif /* cglm_curve_h */

include/cglm/ease.h

@@ -0,0 +1,317 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglm_ease_h
+#define cglm_ease_h
+
+#include "common.h"
+
+CGLM_INLINE
+float
+glm_ease_linear(float t) {
+  return t;
+}
+
+CGLM_INLINE
+float
+glm_ease_sine_in(float t) {
+  return sinf((t - 1.0f) * GLM_PI_2f) + 1.0f;
+}
+
+CGLM_INLINE
+float
+glm_ease_sine_out(float t) {
+  return sinf(t * GLM_PI_2f);
+}
+
+CGLM_INLINE
+float
+glm_ease_sine_inout(float t) {
+  return 0.5f * (1.0f - cosf(t * GLM_PIf));
+}
+
+CGLM_INLINE
+float
+glm_ease_quad_in(float t) {
+  return t * t;
+}
+
+CGLM_INLINE
+float
+glm_ease_quad_out(float t) {
+  return -(t * (t - 2.0f));
+}
+
+CGLM_INLINE
+float
+glm_ease_quad_inout(float t) {
+  float tt;
+
+  tt = t * t;
+  if (t < 0.5f)
+    return 2.0f * tt;
+
+  return (-2.0f * tt) + (4.0f * t) - 1.0f;
+}
+
+CGLM_INLINE
+float
+glm_ease_cubic_in(float t) {
+  return t * t * t;
+}
+
+CGLM_INLINE
+float
+glm_ease_cubic_out(float t) {
+  float f;
+  f = t - 1.0f;
+  return f * f * f + 1.0f;
+}
+
+CGLM_INLINE
+float
+glm_ease_cubic_inout(float t) {
+  float f;
+
+  if (t < 0.5f)
+    return 4.0f * t * t * t;
+
+  f = 2.0f * t - 2.0f;
+
+  return 0.5f * f * f * f + 1.0f;
+}
+
+CGLM_INLINE
+float
+glm_ease_quart_in(float t) {
+  float f;
+  f = t * t;
+  return f * f;
+}
+
+CGLM_INLINE
+float
+glm_ease_quart_out(float t) {
+  float f;
+
+  f = t - 1.0f;
+
+  return f * f * f * (1.0f - t) + 1.0f;
+}
+
+CGLM_INLINE
+float
+glm_ease_quart_inout(float t) {
+  float f, g;
+
+  if (t < 0.5f) {
+    f = t * t;
+    return 8.0f * f * f;
+  }
+
+  f = t - 1.0f;
+  g = f * f;
+
+  return -8.0f * g * g + 1.0f;
+}
+
+CGLM_INLINE
+float
+glm_ease_quint_in(float t) {
+  float f;
+  f = t * t;
+  return f * f * t;
+}
+
+CGLM_INLINE
+float
+glm_ease_quint_out(float t) {
+  float f, g;
+
+  f = t - 1.0f;
+  g = f * f;
+
+  return g * g * f + 1.0f;
+}
+
+CGLM_INLINE
+float
+glm_ease_quint_inout(float t) {
+  float f, g;
+
+  if (t < 0.5f) {
+    f = t * t;
+    return 16.0f * f * f * t;
+  }
+
+  f = 2.0f * t - 2.0f;
+  g = f * f;
+
+  return 0.5f * g * g * f + 1.0f;
+}
+
+CGLM_INLINE
+float
+glm_ease_exp_in(float t) {
+  if (t == 0.0f)
+    return t;
+
+  return powf(2.0f,  10.0f * (t - 1.0f));
+}
+
+CGLM_INLINE
+float
+glm_ease_exp_out(float t) {
+  if (t == 1.0f)
+    return t;
+
+  return 1.0f - powf(2.0f, -10.0f * t);
+}
+
+CGLM_INLINE
+float
+glm_ease_exp_inout(float t) {
+  if (t == 0.0f || t == 1.0f)
+    return t;
+
+  if (t < 0.5f)
+    return 0.5f * powf(2.0f, (20.0f * t) - 10.0f);
+
+  return -0.5f * powf(2.0f, (-20.0f * t) + 10.0f) + 1.0f;
+}
+
+CGLM_INLINE
+float
+glm_ease_circ_in(float t) {
+  return 1.0f - sqrtf(1.0f - (t * t));
+}
+
+CGLM_INLINE
+float
+glm_ease_circ_out(float t) {
+  return sqrtf((2.0f - t) * t);
+}
+
+CGLM_INLINE
+float
+glm_ease_circ_inout(float t) {
+  if (t < 0.5f)
+    return 0.5f * (1.0f - sqrtf(1.0f - 4.0f * (t * t)));
+
+  return 0.5f * (sqrtf(-((2.0f * t) - 3.0f) * ((2.0f * t) - 1.0f)) + 1.0f);
+}
+
+CGLM_INLINE
+float
+glm_ease_back_in(float t) {
+  float o, z;
+
+  o = 1.70158f;
+  z = ((o + 1.0f) * t) - o;
+
+  return t * t * z;
+}
+
+CGLM_INLINE
+float
+glm_ease_back_out(float t) {
+  float o, z, n;
+
+  o = 1.70158f;
+  n = t - 1.0f;
+  z = (o + 1.0f) * n + o;
+
+  return n * n * z + 1.0f;
+}
+
+CGLM_INLINE
+float
+glm_ease_back_inout(float t) {
+  float o, z, n, m, s, x;
+
+  o = 1.70158f;
+  s = o * 1.525f;
+  x = 0.5;
+  n = t / 0.5f;
+
+  if (n < 1.0f) {
+    z = (s + 1) * n - s;
+    m = n * n * z;
+    return x * m;
+  }
+
+  n -= 2.0f;
+  z  = (s + 1.0f) * n + s;
+  m  = (n * n * z) + 2;
+
+  return x * m;
+}
+
+CGLM_INLINE
+float
+glm_ease_elast_in(float t) {
+  return sinf(13.0f * GLM_PI_2f * t) * powf(2.0f, 10.0f * (t - 1.0f));
+}
+
+CGLM_INLINE
+float
+glm_ease_elast_out(float t) {
+  return sinf(-13.0f * GLM_PI_2f * (t + 1.0f)) * powf(2.0f, -10.0f * t) + 1.0f;
+}
+
+CGLM_INLINE
+float
+glm_ease_elast_inout(float t) {
+  float a;
+
+  a = 2.0f * t;
+
+  if (t < 0.5f)
+    return 0.5f * sinf(13.0f * GLM_PI_2f * a)
+                * powf(2.0f, 10.0f * (a - 1.0f));
+
+  return 0.5f * (sinf(-13.0f * GLM_PI_2f * a)
+                 * powf(2.0f, -10.0f * (a - 1.0f)) + 2.0f);
+}
+
+CGLM_INLINE
+float
+glm_ease_bounce_out(float t) {
+  float tt;
+
+  tt = t * t;
+
+  if (t < (4.0f / 11.0f))
+    return (121.0f * tt) / 16.0f;
+
+  if (t < 8.0f / 11.0f)
+    return ((363.0f / 40.0f) * tt) - ((99.0f / 10.0f) * t) + (17.0f / 5.0f);
+
+  if (t < (9.0f / 10.0f))
+    return (4356.0f / 361.0f) * tt
+            - (35442.0f / 1805.0f) * t
+            + (16061.0f / 1805.0f);
+
+  return ((54.0f / 5.0f) * tt) - ((513.0f / 25.0f) * t) + (268.0f / 25.0f);
+}
+
+CGLM_INLINE
+float
+glm_ease_bounce_in(float t) {
+  return 1.0f - glm_ease_bounce_out(1.0f - t);
+}
+
+CGLM_INLINE
+float
+glm_ease_bounce_inout(float t) {
+  if (t < 0.5f)
+    return 0.5f * (1.0f - glm_ease_bounce_out(t * 2.0f));
+
+  return 0.5f * glm_ease_bounce_out(t * 2.0f - 1.0f) + 0.5f;
+}
+
+#endif /* cglm_ease_h */

include/cglm/euler.h

@@ -5,55 +5,202 @@
  * Full license can be found in the LICENSE file
  */
 
+/*
+ NOTE:
+  angles must be passed as [X-Angle, Y-Angle, Z-angle] order
+  For instance you don't pass angles as [Z-Angle, X-Angle, Y-angle] to
+  glm_euler_zxy funciton, All RELATED functions accept angles same order
+  which is [X, Y, Z].
+ */
+
 /*
  Types:
-   enum glm_euler_sq
+   enum glm_euler_seq
 
  Functions:
-   CGLM_INLINE glm_euler_sq glm_euler_order(int newOrder[3]);
-   CGLM_INLINE void glm_euler_angles(mat4 m, vec3 dest);
-   CGLM_INLINE void glm_euler(vec3 angles, mat4 dest);
-   CGLM_INLINE void glm_euler_zyx(vec3 angles, mat4 dest);
-   CGLM_INLINE void glm_euler_zxy(vec3 angles, mat4 dest);
-   CGLM_INLINE void glm_euler_xzy(vec3 angles, mat4 dest);
-   CGLM_INLINE void glm_euler_yzx(vec3 angles, mat4 dest);
-   CGLM_INLINE void glm_euler_yxz(vec3 angles, mat4 dest);
-   CGLM_INLINE void glm_euler_by_order(vec3         angles,
-                                       glm_euler_sq axis,
-                                       mat4         dest);
+   CGLM_INLINE glm_eul_mat4(vec3 ea, int order, mat4 dest)
  */
 
 #ifndef cglm_euler_h
 #define cglm_euler_h
 
 #include "common.h"
+#include "util.h"
+
+/* ---------- Notice for Ken Shoemake's algorithm Implementation -------------*
+ | Ken Shoemake's algorithm impl. is taken from this repo by permission:      |
+ |   https://github.com/erich666/GraphicsGems/blob/master/gemsiv/euler_angle  |
+ |                                                                            |
+ | cglm doesn't claim the ownership of GraphicsGems source codes              |
+ | and the algorithm itself. But cglm may change variable names or some piece |
+ | of codes in order to apply optimizations or to make it usable in cglm.     |
+ |                                                                            |
+ | Related issue: https://github.com/recp/cglm/issues/30                      |
+ |                                                                            |
+ * -------------------------- GraphicsGems EULA ----------------------------- *
+ | Related EULA for GraphicsGems can be found at below, plus in CREDITS:      |
+ |   http://www.realtimerendering.com/resources/GraphicsGems/                 |
+ |                                                                            |
+ | EULA: The Graphics Gems code is copyright-protected. In other words, you   |
+ | cannot claim the text of the code as your own and resell it. Using the     |
+ | code is permitted in any program, product, or library, non-commercial or   |
+ | commercial. Giving credit is not required, though is a nice gesture.       |
+ | The code comes as-is, and if there are any flaws or problems with any Gems |
+ | code, nobody involved with Gems - authors, editors, publishers, or         |
+ | webmasters - are to be held responsible. Basically, don't be a jerk, and   |
+ | remember that anything free comes with no guarantee.                       |
+ * -------------------------------- END --------------------------------------*/
+
+/* Order type constants, constructors, extractors
+ * There are 24 possible conventions, designated by:
+ *	  o EulAxI = axis used initially
+ *	  o EulPar = parity of axis permutation
+ *	  o EulRep = repetition of initial axis as last
+ *	  o EulFrm = frame from which axes are taken
+ * Axes I,J,K will be a permutation of X,Y,Z.
+ * Axis H will be either I or K, depending on EulRep.
+ * Frame S takes axes from initial static frame.
+ * If ord = (AxI=X, Par=Even, Rep=No, Frm=S), then
+ * {a,b,c,ord} means Rz(c)Ry(b)Rx(a), where Rz(c)v
+ * rotates v around Z by c radians.			    
+*/
+
+#define EulRep(ord)  (((unsigned)(ord)>>1)&1)
+#define EulFrm(ord)  ((unsigned)(ord)&1)
+#define EulPar(ord)  (((unsigned)(ord)>>2)&1)
+
+/*! this code is merely a quick (and legal!) way to set arrays,
+    EulSafe being 0,1,2,0 */
+#define EulSafe	     "\000\001\002\000"
+#define EulNext	     "\001\002\000\001"
+#define EulAxI(ord)  ((int)(EulSafe[(((unsigned)(ord)>>3)&3)]))
+#define EulAxJ(ord)  ((int)(EulNext[EulAxI(ord)+(EulPar(ord)==EulParOdd)]))
+#define EulAxK(ord)  ((int)(EulNext[EulAxI(ord)+(EulPar(ord)!=EulParOdd)]))
+#define EulAxH(ord)  ((EulRep(ord)==EulRepNo)?EulAxK(ord):EulAxI(ord))
+
+/*! EulGetOrd unpacks all useful information about order simultaneously. */
+#define EulGetOrd(ord,i,j,k,h,n,s,f)                                          \
+  {unsigned o=(unsigned)ord;f=o&1;o>>=1;s=o&1;o>>=1;\
+    n=o&1;o>>=1;i=EulSafe[o&3];j=EulNext[i+n];k=EulNext[i+1-n];h=s?k:i;}
+
+typedef enum glm_eul_order {
+  /*! Static axes */
+  GLM_EUL_XYZs  = 0,
+  GLM_EUL_XYXs  = 2,
+  GLM_EUL_XZYs  = 4,
+  GLM_EUL_XZXs  = 6,
+  GLM_EUL_YZXs  = 8,
+  GLM_EUL_YZYs  = 10,
+  GLM_EUL_YXZs  = 12,
+  GLM_EUL_YXYs  = 14,
+  GLM_EUL_ZXYs  = 16,
+  GLM_EUL_ZXZs  = 18,
+  GLM_EUL_ZYXs  = 20,
+  GLM_EUL_ZYZs  = 22,
+  
+  /*! Rotating axes */
+  GLM_EUL_ZYXr  = 1,
+  GLM_EUL_XYXr  = 3,
+  GLM_EUL_YZXr  = 5,
+  GLM_EUL_XZXr  = 7,
+  GLM_EUL_XZYr  = 9,
+  GLM_EUL_YZYr  = 11,
+  GLM_EUL_ZXYr  = 13,
+  GLM_EUL_YXYr  = 15,
+  GLM_EUL_YXZr  = 17,
+  GLM_EUL_ZXZr  = 19,
+  GLM_EUL_XYZr  = 21,
+  GLM_EUL_ZYZr  = 23
+} glm_eul_order;
+
+/*!
+ * @brief build matrix from euler angles
+ *
+ * @param[in]  ea     [Xangle, Yangle, Zangle]
+ * @param[out] dest   rotation matrix
+ */
+CGLM_INLINE
+void
+glm_eul_mat4(vec3 ea, glm_eul_order order, mat4 dest) {
+  float ti, tj, th, ci, cj, ch, si, sj, sh, cc, cs, sc, ss;
+  int   i, j, k, h, parOdd, repYes, frmR;
+
+  EulGetOrd(order, i, j, k, h, parOdd, repYes, frmR);
+  
+  if (frmR == 1)
+    glm_swapf(&ea[0], &ea[2]);
+  
+  if (parOdd == 1)
+    glm_vec3_negate(ea);
+  
+  ti = ea[0]; tj = ea[1]; th = ea[2];
+  
+  ci = cosf(ti);  cj = cosf(tj);
+  ch = cosf(th);  si = sinf(ti);
+  sj = sinf(tj);  sh = sinf(th);
+  
+  cc = ci * ch;  cs = ci * sh;
+  sc = si * ch;  ss = si * sh;
+  
+  if (repYes == 1) {
+    dest[i][i] =  cj;
+    dest[i][j] =  sj * si;
+    dest[i][k] =  sj * ci;
+    dest[j][i] =  sj * sh;
+    dest[j][j] = -cj * ss + cc;
+    dest[j][k] = -cj * cs - sc;
+    dest[k][i] = -sj * ch;
+    dest[k][j] =  cj * sc + cs;
+    dest[k][k] =  cj * cc - ss;
+  } else {
+    dest[i][i] =  cj * ch;
+    dest[i][j] =  sj * sc - cs;
+    dest[i][k] =  sj * cc + ss;
+    dest[j][i] =  cj * sh;
+    dest[j][j] =  sj * ss + cc;
+    dest[j][k] =  sj * cs - sc;
+    dest[k][i] = -sj;
+    dest[k][j] =  cj * si;
+    dest[k][k] =  cj * ci;
+  }
+  
+  dest[3][0] = 0.f;
+  dest[3][1] = 0.f;
+  dest[3][2] = 0.f;
+  dest[0][3] = 0.f;
+  dest[1][3] = 0.f;
+  dest[2][3] = 0.f;
+  dest[3][3] = 1.f;
+}
 
 /*!
  * if you have axis order like vec3 orderVec = [0, 1, 2] or [0, 2, 1]...
  * vector then you can convert it to this enum by doing this:
  * @code
- * glm_euler_sq order;
+ * glm_euler_seq order;
  * order = orderVec[0] | orderVec[1] << 2 | orderVec[2] << 4;
  * @endcode
  * you may need to explicit cast if required
  */
-typedef enum glm_euler_sq {
+typedef enum glm_euler_seq {
   GLM_EULER_XYZ = 0 << 0 | 1 << 2 | 2 << 4,
   GLM_EULER_XZY = 0 << 0 | 2 << 2 | 1 << 4,
   GLM_EULER_YZX = 1 << 0 | 2 << 2 | 0 << 4,
   GLM_EULER_YXZ = 1 << 0 | 0 << 2 | 2 << 4,
   GLM_EULER_ZXY = 2 << 0 | 0 << 2 | 1 << 4,
   GLM_EULER_ZYX = 2 << 0 | 1 << 2 | 0 << 4
-} glm_euler_sq;
+} glm_euler_seq;
+
+typedef glm_euler_seq glm_euler_sq;
 
 CGLM_INLINE
-glm_euler_sq
-glm_euler_order(int newOrder[3]) {
-  return (glm_euler_sq)(newOrder[0] | newOrder[1] << 2 | newOrder[2] << 4);
+glm_euler_seq
+glm_euler_order(int ord[3]) {
+  return (glm_euler_seq)(ord[0] << 0 | ord[1] << 2 | ord[2] << 4);
 }
 
 /*!
- * @brief euler angles (in radian) using xyz sequence
+ * @brief extract euler angles (in radians) using xyz order
  *
  * @param[in]  m    affine transform
  * @param[out] dest angles vector [x, y, z]
@@ -61,225 +208,289 @@ glm_euler_order(int newOrder[3]) {
 CGLM_INLINE
 void
 glm_euler_angles(mat4 m, vec3 dest) {
-  if (m[0][2] < 1.0f) {
-    if (m[0][2] > -1.0f) {
-      vec3  a[2];
-      float cy1, cy2;
-      int   path;
+  float m00, m01, m10, m11, m20, m21, m22;
+  float thetaX, thetaY, thetaZ;
 
-      a[0][1] = asinf(-m[0][2]);
-      a[1][1] = CGLM_PI - a[0][1];
+  m00 = m[0][0];  m10 = m[1][0];  m20 = m[2][0];
+  m01 = m[0][1];  m11 = m[1][1];  m21 = m[2][1];
+                                  m22 = m[2][2];
 
-      cy1 = cosf(a[0][1]);
-      cy2 = cosf(a[1][1]);
-
-      a[0][0] = atan2f(m[1][2] / cy1, m[2][2] / cy1);
-      a[1][0] = atan2f(m[1][2] / cy2, m[2][2] / cy2);
-
-      a[0][2] = atan2f(m[0][1] / cy1, m[0][0] / cy1);
-      a[1][2] = atan2f(m[0][1] / cy2, m[0][0] / cy2);
-
-      path = (fabsf(a[0][0]) + fabsf(a[0][1]) + fabsf(a[0][2])) >=
-               (fabsf(a[1][0]) + fabsf(a[1][1]) + fabsf(a[1][2]));
-
-      glm_vec_copy(a[path], dest);
-    } else {
-      dest[0] = atan2f(m[1][0], m[2][0]);
-      dest[1] = CGLM_PI_2;
-      dest[2] = 0.0f;
+  if (m20 < 1.0f) {
+    if (m20 > -1.0f) {
+      thetaY = asinf(m20);
+      thetaX = atan2f(-m21, m22);
+      thetaZ = atan2f(-m10, m00);
+    } else { /* m20 == -1 */
+      /* Not a unique solution */
+      thetaY = -GLM_PI_2f;
+      thetaX = -atan2f(m01, m11);
+      thetaZ =  0.0f;
     }
-  } else {
-    dest[0] = atan2f(-m[1][0], -m[2][0]);
-    dest[1] =-CGLM_PI_2;
-    dest[2] = 0.0f;
+  } else { /* m20 == +1 */
+    thetaY = GLM_PI_2f;
+    thetaX = atan2f(m01, m11);
+    thetaZ = 0.0f;
   }
+
+  dest[0] = thetaX;
+  dest[1] = thetaY;
+  dest[2] = thetaZ;
 }
 
 /*!
- * @brief build rotation matrix from euler angles(ExEyEz/RzRyRx)
+ * @brief build rotation matrix from euler angles
  *
- * @param[in]  angles angles as vector [Ex, Ey, Ez]
+ * @param[in]  angles angles as vector [Xangle, Yangle, Zangle]
+ * @param[out] dest   rotation matrix
+ */
+CGLM_INLINE
+void
+glm_euler_xyz(vec3 angles, mat4 dest) {
+  float cx, cy, cz,
+        sx, sy, sz, czsx, cxcz, sysz;
+
+  sx   = sinf(angles[0]); cx = cosf(angles[0]);
+  sy   = sinf(angles[1]); cy = cosf(angles[1]);
+  sz   = sinf(angles[2]); cz = cosf(angles[2]);
+
+  czsx = cz * sx;
+  cxcz = cx * cz;
+  sysz = sy * sz;
+
+  dest[0][0] =  cy * cz;
+  dest[0][1] =  czsx * sy + cx * sz;
+  dest[0][2] = -cxcz * sy + sx * sz;
+  dest[1][0] = -cy * sz;
+  dest[1][1] =  cxcz - sx * sysz;
+  dest[1][2] =  czsx + cx * sysz;
+  dest[2][0] =  sy;
+  dest[2][1] = -cy * sx;
+  dest[2][2] =  cx * cy;
+  dest[0][3] =  0.0f;
+  dest[1][3] =  0.0f;
+  dest[2][3] =  0.0f;
+  dest[3][0] =  0.0f;
+  dest[3][1] =  0.0f;
+  dest[3][2] =  0.0f;
+  dest[3][3] =  1.0f;
+}
+
+/*!
+ * @brief build rotation matrix from euler angles
+ *
+ * @param[in]  angles angles as vector [Xangle, Yangle, Zangle]
  * @param[out] dest   rotation matrix
  */
 CGLM_INLINE
 void
 glm_euler(vec3 angles, mat4 dest) {
-  float cx, cy, cz,
-        sx, sy, sz;
-
-  sx = sinf(angles[0]); cx = cosf(angles[0]);
-  sy = sinf(angles[1]); cy = cosf(angles[1]);
-  sz = sinf(angles[2]); cz = cosf(angles[2]);
-
-  dest[0][0] = cy * cz;
-  dest[0][1] = cy * sz;
-  dest[0][2] =-sy;
-  dest[1][0] = cz * sx * sy - cx * sz;
-  dest[1][1] = cx * cz + sx * sy * sz;
-  dest[1][2] = cy * sx;
-  dest[2][0] = cx * cz * sy + sx * sz;
-  dest[2][1] =-cz * sx + cx * sy * sz;
-  dest[2][2] = cx * cy;
-  dest[0][3] = 0.0f;
-  dest[1][3] = 0.0f;
-  dest[2][3] = 0.0f;
-  dest[3][0] = 0.0f;
-  dest[3][1] = 0.0f;
-  dest[3][2] = 0.0f;
-  dest[3][3] = 1.0f;
+  glm_euler_xyz(angles, dest);
 }
 
 /*!
- * @brief build rotation matrix from euler angles (EzEyEx/RxRyRz)
+ * @brief build rotation matrix from euler angles
+ *
+ * @param[in]  angles angles as vector [Xangle, Yangle, Zangle]
+ * @param[out] dest   rotation matrix
  */
 CGLM_INLINE
 void
-glm_euler_zyx(vec3 angles,
-              mat4 dest) {
+glm_euler_xzy(vec3 angles, mat4 dest) {
   float cx, cy, cz,
-        sx, sy, sz;
+  sx, sy, sz, sxsy, cysx, cxsy, cxcy;
 
-  sx = sinf(angles[0]); cx = cosf(angles[0]);
-  sy = sinf(angles[1]); cy = cosf(angles[1]);
-  sz = sinf(angles[2]); cz = cosf(angles[2]);
+  sx   = sinf(angles[0]); cx = cosf(angles[0]);
+  sy   = sinf(angles[1]); cy = cosf(angles[1]);
+  sz   = sinf(angles[2]); cz = cosf(angles[2]);
 
-  dest[0][0] = cy * cz;
-  dest[0][1] = cz * sx * sy + cx * sz;
-  dest[0][2] =-cx * cz * sy + sx * sz;
-  dest[1][0] =-cy * sz;
-  dest[1][1] = cx * cz - sx * sy * sz;
-  dest[1][2] = cz * sx + cx * sy * sz;
-  dest[2][0] = sy;
-  dest[2][1] =-cy * sx;
-  dest[2][2] = cx * cy;
-  dest[0][3] = 0.0f;
-  dest[1][3] = 0.0f;
-  dest[2][3] = 0.0f;
-  dest[3][0] = 0.0f;
-  dest[3][1] = 0.0f;
-  dest[3][2] = 0.0f;
-  dest[3][3] = 1.0f;
+  sxsy = sx * sy;
+  cysx = cy * sx;
+  cxsy = cx * sy;
+  cxcy = cx * cy;
+
+  dest[0][0] =  cy * cz;
+  dest[0][1] =  sxsy + cxcy * sz;
+  dest[0][2] = -cxsy + cysx * sz;
+  dest[1][0] = -sz;
+  dest[1][1] =  cx * cz;
+  dest[1][2] =  cz * sx;
+  dest[2][0] =  cz * sy;
+  dest[2][1] = -cysx + cxsy * sz;
+  dest[2][2] =  cxcy + sxsy * sz;
+  dest[0][3] =  0.0f;
+  dest[1][3] =  0.0f;
+  dest[2][3] =  0.0f;
+  dest[3][0] =  0.0f;
+  dest[3][1] =  0.0f;
+  dest[3][2] =  0.0f;
+  dest[3][3] =  1.0f;
 }
 
+
+/*!
+ * @brief build rotation matrix from euler angles
+ *
+ * @param[in]  angles angles as vector [Xangle, Yangle, Zangle]
+ * @param[out] dest   rotation matrix
+ */
 CGLM_INLINE
 void
-glm_euler_zxy(vec3 angles,
-              mat4 dest) {
+glm_euler_yxz(vec3 angles, mat4 dest) {
   float cx, cy, cz,
-        sx, sy, sz;
+        sx, sy, sz, cycz, sysz, czsy, cysz;
 
-  sx = sinf(angles[0]); cx = cosf(angles[0]);
-  sy = sinf(angles[1]); cy = cosf(angles[1]);
-  sz = sinf(angles[2]); cz = cosf(angles[2]);
+  sx   = sinf(angles[0]); cx = cosf(angles[0]);
+  sy   = sinf(angles[1]); cy = cosf(angles[1]);
+  sz   = sinf(angles[2]); cz = cosf(angles[2]);
 
-  dest[0][0] = cy * cz + sx * sy * sz;
-  dest[0][1] = cx * sz;
-  dest[0][2] =-cz * sy + cy * sx * sz;
-  dest[1][0] = cz * sx * sy - cy * sz;
-  dest[1][1] = cx * cz;
-  dest[1][2] = cy * cz * sx + sy * sz;
-  dest[2][0] = cx * sy;
-  dest[2][1] =-sx;
-  dest[2][2] = cx * cy;
-  dest[0][3] = 0.0f;
-  dest[1][3] = 0.0f;
-  dest[2][3] = 0.0f;
-  dest[3][0] = 0.0f;
-  dest[3][1] = 0.0f;
-  dest[3][2] = 0.0f;
-  dest[3][3] = 1.0f;
+  cycz = cy * cz;
+  sysz = sy * sz;
+  czsy = cz * sy;
+  cysz = cy * sz;
+
+  dest[0][0] =  cycz + sx * sysz;
+  dest[0][1] =  cx * sz;
+  dest[0][2] = -czsy + cysz * sx;
+  dest[1][0] = -cysz + czsy * sx;
+  dest[1][1] =  cx * cz;
+  dest[1][2] =  cycz * sx + sysz;
+  dest[2][0] =  cx * sy;
+  dest[2][1] = -sx;
+  dest[2][2] =  cx * cy;
+  dest[0][3] =  0.0f;
+  dest[1][3] =  0.0f;
+  dest[2][3] =  0.0f;
+  dest[3][0] =  0.0f;
+  dest[3][1] =  0.0f;
+  dest[3][2] =  0.0f;
+  dest[3][3] =  1.0f;
 }
 
+/*!
+ * @brief build rotation matrix from euler angles
+ *
+ * @param[in]  angles angles as vector [Xangle, Yangle, Zangle]
+ * @param[out] dest   rotation matrix
+ */
 CGLM_INLINE
 void
-glm_euler_xzy(vec3 angles,
-              mat4 dest) {
+glm_euler_yzx(vec3 angles, mat4 dest) {
   float cx, cy, cz,
-        sx, sy, sz;
+        sx, sy, sz, sxsy, cxcy, cysx, cxsy;
 
-  sx = sinf(angles[0]); cx = cosf(angles[0]);
-  sy = sinf(angles[1]); cy = cosf(angles[1]);
-  sz = sinf(angles[2]); cz = cosf(angles[2]);
+  sx   = sinf(angles[0]); cx = cosf(angles[0]);
+  sy   = sinf(angles[1]); cy = cosf(angles[1]);
+  sz   = sinf(angles[2]); cz = cosf(angles[2]);
 
-  dest[0][0] = cy * cz;
-  dest[0][1] = sz;
-  dest[0][2] =-cz * sy;
-  dest[1][0] = sx * sy - cx * cy * sz;
-  dest[1][1] = cx * cz;
-  dest[1][2] = cy * sx + cx * sy * sz;
-  dest[2][0] = cx * sy + cy * sx * sz;
-  dest[2][1] =-cz * sx;
-  dest[2][2] = cx * cy - sx * sy * sz;
-  dest[0][3] = 0.0f;
-  dest[1][3] = 0.0f;
-  dest[2][3] = 0.0f;
-  dest[3][0] = 0.0f;
-  dest[3][1] = 0.0f;
-  dest[3][2] = 0.0f;
-  dest[3][3] = 1.0f;
+  sxsy = sx * sy;
+  cxcy = cx * cy;
+  cysx = cy * sx;
+  cxsy = cx * sy;
+
+  dest[0][0] =  cy * cz;
+  dest[0][1] =  sz;
+  dest[0][2] = -cz * sy;
+  dest[1][0] =  sxsy - cxcy * sz;
+  dest[1][1] =  cx * cz;
+  dest[1][2] =  cysx + cxsy * sz;
+  dest[2][0] =  cxsy + cysx * sz;
+  dest[2][1] = -cz * sx;
+  dest[2][2] =  cxcy - sxsy * sz;
+  dest[0][3] =  0.0f;
+  dest[1][3] =  0.0f;
+  dest[2][3] =  0.0f;
+  dest[3][0] =  0.0f;
+  dest[3][1] =  0.0f;
+  dest[3][2] =  0.0f;
+  dest[3][3] =  1.0f;
 }
 
+/*!
+ * @brief build rotation matrix from euler angles
+ *
+ * @param[in]  angles angles as vector [Xangle, Yangle, Zangle]
+ * @param[out] dest   rotation matrix
+ */
 CGLM_INLINE
 void
-glm_euler_yzx(vec3 angles,
-              mat4 dest) {
+glm_euler_zxy(vec3 angles, mat4 dest) {
   float cx, cy, cz,
-        sx, sy, sz;
+        sx, sy, sz, cycz, sxsy, cysz;
 
-  sx = sinf(angles[0]); cx = cosf(angles[0]);
-  sy = sinf(angles[1]); cy = cosf(angles[1]);
-  sz = sinf(angles[2]); cz = cosf(angles[2]);
+  sx   = sinf(angles[0]); cx = cosf(angles[0]);
+  sy   = sinf(angles[1]); cy = cosf(angles[1]);
+  sz   = sinf(angles[2]); cz = cosf(angles[2]);
 
-  dest[0][0] = cy * cz;
-  dest[0][1] = sx * sy + cx * cy * sz;
-  dest[0][2] =-cx * sy + cy * sx * sz;
-  dest[1][0] =-sz;
-  dest[1][1] = cx * cz;
-  dest[1][2] = cz * sx;
-  dest[2][0] = cz * sy;
-  dest[2][1] =-cy * sx + cx * sy * sz;
-  dest[2][2] = cx * cy + sx * sy * sz;
-  dest[0][3] = 0.0f;
-  dest[1][3] = 0.0f;
-  dest[2][3] = 0.0f;
-  dest[3][0] = 0.0f;
-  dest[3][1] = 0.0f;
-  dest[3][2] = 0.0f;
-  dest[3][3] = 1.0f;
+  cycz = cy * cz;
+  sxsy = sx * sy;
+  cysz = cy * sz;
+
+  dest[0][0] =  cycz - sxsy * sz;
+  dest[0][1] =  cz * sxsy + cysz;
+  dest[0][2] = -cx * sy;
+  dest[1][0] = -cx * sz;
+  dest[1][1] =  cx * cz;
+  dest[1][2] =  sx;
+  dest[2][0] =  cz * sy + cysz * sx;
+  dest[2][1] = -cycz * sx + sy * sz;
+  dest[2][2] =  cx * cy;
+  dest[0][3] =  0.0f;
+  dest[1][3] =  0.0f;
+  dest[2][3] =  0.0f;
+  dest[3][0] =  0.0f;
+  dest[3][1] =  0.0f;
+  dest[3][2] =  0.0f;
+  dest[3][3] =  1.0f;
 }
 
+/*!
+ * @brief build rotation matrix from euler angles
+ *
+ * @param[in]  angles angles as vector [Xangle, Yangle, Zangle]
+ * @param[out] dest   rotation matrix
+ */
 CGLM_INLINE
 void
-glm_euler_yxz(vec3 angles,
-              mat4 dest) {
+glm_euler_zyx(vec3 angles, mat4 dest) {
   float cx, cy, cz,
-        sx, sy, sz;
+        sx, sy, sz, czsx, cxcz, sysz;
 
-  sx = sinf(angles[0]); cx = cosf(angles[0]);
-  sy = sinf(angles[1]); cy = cosf(angles[1]);
-  sz = sinf(angles[2]); cz = cosf(angles[2]);
+  sx   = sinf(angles[0]); cx = cosf(angles[0]);
+  sy   = sinf(angles[1]); cy = cosf(angles[1]);
+  sz   = sinf(angles[2]); cz = cosf(angles[2]);
 
-  dest[0][0] = cy * cz - sx * sy * sz;
-  dest[0][1] = cz * sx * sy + cy * sz;
-  dest[0][2] =-cx * sy;
-  dest[1][0] =-cx * sz;
-  dest[1][1] = cx * cz;
-  dest[1][2] = sx;
-  dest[2][0] = cz * sy + cy * sx * sz;
-  dest[2][1] =-cy * cz * sx + sy * sz;
-  dest[2][2] = cx * cy;
-  dest[0][3] = 0.0f;
-  dest[1][3] = 0.0f;
-  dest[2][3] = 0.0f;
-  dest[3][0] = 0.0f;
-  dest[3][1] = 0.0f;
-  dest[3][2] = 0.0f;
-  dest[3][3] = 1.0f;
+  czsx = cz * sx;
+  cxcz = cx * cz;
+  sysz = sy * sz;
+
+  dest[0][0] =  cy * cz;
+  dest[0][1] =  cy * sz;
+  dest[0][2] = -sy;
+  dest[1][0] =  czsx * sy - cx * sz;
+  dest[1][1] =  cxcz + sx * sysz;
+  dest[1][2] =  cy * sx;
+  dest[2][0] =  cxcz * sy + sx * sz;
+  dest[2][1] = -czsx + cx * sysz;
+  dest[2][2] =  cx * cy;
+  dest[0][3] =  0.0f;
+  dest[1][3] =  0.0f;
+  dest[2][3] =  0.0f;
+  dest[3][0] =  0.0f;
+  dest[3][1] =  0.0f;
+  dest[3][2] =  0.0f;
+  dest[3][3] =  1.0f;
 }
 
+/*!
+ * @brief build rotation matrix from euler angles
+ *
+ * @param[in]  angles angles as vector [Xangle, Yangle, Zangle]
+ * @param[in]  ord    euler order
+ * @param[out] dest   rotation matrix
+ */
 CGLM_INLINE
 void
-glm_euler_by_order(vec3 angles, glm_euler_sq axis, mat4 dest) {
+glm_euler_by_order(vec3 angles, glm_euler_seq ord, mat4 dest) {
   float cx, cy, cz,
         sx, sy, sz;
 
@@ -297,72 +508,72 @@ glm_euler_by_order(vec3 angles, glm_euler_sq axis, mat4 dest) {
   czsx = cz * sx; cxsz = cx * sz;
   sysz = sy * sz;
 
-  switch (axis) {
-    case GLM_EULER_XYZ:
-      dest[0][0] = cycz;
-      dest[0][1] = cysz;
-      dest[0][2] =-sy;
-      dest[1][0] = czsx * sy - cxsz;
-      dest[1][1] = cxcz + sx * sysz;
-      dest[1][2] = cysx;
-      dest[2][0] = cx * czsy + sx * sz;
-      dest[2][1] =-czsx + cx * sysz;
-      dest[2][2] = cxcy;
-      break;
+  switch (ord) {
     case GLM_EULER_XZY:
-      dest[0][0] = cycz;
-      dest[0][1] = sz;
-      dest[0][2] =-czsy;
-      dest[1][0] = sx * sy - cx * cysz;
-      dest[1][1] = cxcz;
-      dest[1][2] = cysx + cx * sysz;
-      dest[2][0] = cx * sy + cysx * sz;
-      dest[2][1] =-czsx;
-      dest[2][2] = cxcy - sx * sysz;
+      dest[0][0] =  cycz;
+      dest[0][1] =  sx * sy + cx * cysz;
+      dest[0][2] = -cx * sy + cysx * sz;
+      dest[1][0] = -sz;
+      dest[1][1] =  cxcz;
+      dest[1][2] =  czsx;
+      dest[2][0] =  czsy;
+      dest[2][1] = -cysx + cx * sysz;
+      dest[2][2] =  cxcy + sx * sysz;
       break;
-    case GLM_EULER_ZXY:
-      dest[0][0] = cycz + sx * sysz;
-      dest[0][1] = cxsz;
-      dest[0][2] =-czsy + cysx * sz;
-      dest[1][0] = czsx * sy - cysz;
-      dest[1][1] = cxcz;
-      dest[1][2] = cycz * sx + sysz;
-      dest[2][0] = cx * sy;
-      dest[2][1] =-sx;
-      dest[2][2] = cxcy;
-      break;
-    case GLM_EULER_ZYX:
-      dest[0][0] = cycz;
-      dest[0][1] = czsx * sy + cxsz;
-      dest[0][2] =-cx * czsy + sx * sz;
-      dest[1][0] =-cysz;
-      dest[1][1] = cxcz - sx * sysz;
-      dest[1][2] = czsx + cx * sysz;
-      dest[2][0] = sy;
-      dest[2][1] =-cysx;
-      dest[2][2] = cxcy;
+    case GLM_EULER_XYZ:
+      dest[0][0] =  cycz;
+      dest[0][1] =  czsx * sy + cxsz;
+      dest[0][2] = -cx * czsy + sx * sz;
+      dest[1][0] = -cysz;
+      dest[1][1] =  cxcz - sx * sysz;
+      dest[1][2] =  czsx + cx * sysz;
+      dest[2][0] =  sy;
+      dest[2][1] = -cysx;
+      dest[2][2] =  cxcy;
       break;
     case GLM_EULER_YXZ:
-      dest[0][0] = cycz - sx * sysz;
-      dest[0][1] = czsx * sy + cysz;
-      dest[0][2] =-cx * sy;
-      dest[1][0] =-cxsz;
-      dest[1][1] = cxcz;
-      dest[1][2] = sx;
-      dest[2][0] = czsy + cysx * sz;
-      dest[2][1] =-cycz * sx + sysz;
-      dest[2][2] = cxcy;
+      dest[0][0] =  cycz + sx * sysz;
+      dest[0][1] =  cxsz;
+      dest[0][2] = -czsy + cysx * sz;
+      dest[1][0] =  czsx * sy - cysz;
+      dest[1][1] =  cxcz;
+      dest[1][2] =  cycz * sx + sysz;
+      dest[2][0] =  cx * sy;
+      dest[2][1] = -sx;
+      dest[2][2] =  cxcy;
       break;
     case GLM_EULER_YZX:
-      dest[0][0] = cycz;
-      dest[0][1] = sx * sy + cx * cysz;
-      dest[0][2] =-cx * sy + cysx * sz;
-      dest[1][0] =-sz;
-      dest[1][1] = cxcz;
-      dest[1][2] = czsx;
-      dest[2][0] = czsy;
-      dest[2][1] =-cysx + cx * sysz;
-      dest[2][2] = cxcy + sx * sysz;
+      dest[0][0] =  cycz;
+      dest[0][1] =  sz;
+      dest[0][2] = -czsy;
+      dest[1][0] =  sx * sy - cx * cysz;
+      dest[1][1] =  cxcz;
+      dest[1][2] =  cysx + cx * sysz;
+      dest[2][0] =  cx * sy + cysx * sz;
+      dest[2][1] = -czsx;
+      dest[2][2] =  cxcy - sx * sysz;
+      break;
+    case GLM_EULER_ZXY:
+      dest[0][0] =  cycz - sx * sysz;
+      dest[0][1] =  czsx * sy + cysz;
+      dest[0][2] = -cx * sy;
+      dest[1][0] = -cxsz;
+      dest[1][1] =  cxcz;
+      dest[1][2] =  sx;
+      dest[2][0] =  czsy + cysx * sz;
+      dest[2][1] = -cycz * sx + sysz;
+      dest[2][2] =  cxcy;
+      break;
+    case GLM_EULER_ZYX:
+      dest[0][0] =  cycz;
+      dest[0][1] =  cysz;
+      dest[0][2] = -sy;
+      dest[1][0] =  czsx * sy - cxsz;
+      dest[1][1] =  cxcz + sx * sysz;
+      dest[1][2] =  cysx;
+      dest[2][0] =  cx * czsy + sx * sz;
+      dest[2][1] = -czsx + cx * sysz;
+      dest[2][2] =  cxcy;
       break;
   }
 

include/cglm/frustum.h

@@ -0,0 +1,255 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglm_frustum_h
+#define cglm_frustum_h
+
+#include "common.h"
+#include "plane.h"
+#include "vec3.h"
+#include "vec4.h"
+#include "mat4.h"
+
+#define GLM_LBN 0 /* left  bottom near */
+#define GLM_LTN 1 /* left  top    near */
+#define GLM_RTN 2 /* right top    near */
+#define GLM_RBN 3 /* right bottom near */
+
+#define GLM_LBF 4 /* left  bottom far  */
+#define GLM_LTF 5 /* left  top    far  */
+#define GLM_RTF 6 /* right top    far  */
+#define GLM_RBF 7 /* right bottom far  */
+
+#define GLM_LEFT   0
+#define GLM_RIGHT  1
+#define GLM_BOTTOM 2
+#define GLM_TOP    3
+#define GLM_NEAR   4
+#define GLM_FAR    5
+
+/* you can override clip space coords
+   but you have to provide all with same name
+   e.g.: define GLM_CSCOORD_LBN {0.0f, 0.0f, 1.0f, 1.0f} */
+#ifndef GLM_CUSTOM_CLIPSPACE
+
+/* near */
+#define GLM_CSCOORD_LBN {-1.0f, -1.0f, -1.0f, 1.0f}
+#define GLM_CSCOORD_LTN {-1.0f,  1.0f, -1.0f, 1.0f}
+#define GLM_CSCOORD_RTN { 1.0f,  1.0f, -1.0f, 1.0f}
+#define GLM_CSCOORD_RBN { 1.0f, -1.0f, -1.0f, 1.0f}
+
+/* far */
+#define GLM_CSCOORD_LBF {-1.0f, -1.0f,  1.0f, 1.0f}
+#define GLM_CSCOORD_LTF {-1.0f,  1.0f,  1.0f, 1.0f}
+#define GLM_CSCOORD_RTF { 1.0f,  1.0f,  1.0f, 1.0f}
+#define GLM_CSCOORD_RBF { 1.0f, -1.0f,  1.0f, 1.0f}
+
+#endif
+
+/*!
+ * @brief extracts view frustum planes
+ *
+ * planes' space:
+ *  1- if m = proj:     View Space
+ *  2- if m = viewProj: World Space
+ *  3- if m = MVP:      Object Space
+ *
+ * You probably want to extract planes in world space so use viewProj as m
+ * Computing viewProj:
+ *   glm_mat4_mul(proj, view, viewProj);
+ *
+ * Exracted planes order: [left, right, bottom, top, near, far]
+ *
+ * @param[in]  m    matrix (see brief)
+ * @param[out] dest extracted view frustum planes (see brief)
+ */
+CGLM_INLINE
+void
+glm_frustum_planes(mat4 m, vec4 dest[6]) {
+  mat4 t;
+
+  glm_mat4_transpose_to(m, t);
+
+  glm_vec4_add(t[3], t[0], dest[0]); /* left   */
+  glm_vec4_sub(t[3], t[0], dest[1]); /* right  */
+  glm_vec4_add(t[3], t[1], dest[2]); /* bottom */
+  glm_vec4_sub(t[3], t[1], dest[3]); /* top    */
+  glm_vec4_add(t[3], t[2], dest[4]); /* near   */
+  glm_vec4_sub(t[3], t[2], dest[5]); /* far    */
+
+  glm_plane_normalize(dest[0]);
+  glm_plane_normalize(dest[1]);
+  glm_plane_normalize(dest[2]);
+  glm_plane_normalize(dest[3]);
+  glm_plane_normalize(dest[4]);
+  glm_plane_normalize(dest[5]);
+}
+
+/*!
+ * @brief extracts view frustum corners using clip-space coordinates
+ *
+ * corners' space:
+ *  1- if m = invViewProj: World Space
+ *  2- if m = invMVP:      Object Space
+ *
+ * You probably want to extract corners in world space so use invViewProj
+ * Computing invViewProj:
+ *   glm_mat4_mul(proj, view, viewProj);
+ *   ...
+ *   glm_mat4_inv(viewProj, invViewProj);
+ *
+ * if you have a near coord at i index, you can get it's far coord by i + 4
+ *
+ * Find center coordinates:
+ *   for (j = 0; j < 4; j++) {
+ *     glm_vec3_center(corners[i], corners[i + 4], centerCorners[i]);
+ *   }
+ *
+ * @param[in]  invMat matrix (see brief)
+ * @param[out] dest   exracted view frustum corners (see brief)
+ */
+CGLM_INLINE
+void
+glm_frustum_corners(mat4 invMat, vec4 dest[8]) {
+  vec4 c[8];
+
+  /* indexOf(nearCoord) = indexOf(farCoord) + 4 */
+  vec4 csCoords[8] = {
+    GLM_CSCOORD_LBN,
+    GLM_CSCOORD_LTN,
+    GLM_CSCOORD_RTN,
+    GLM_CSCOORD_RBN,
+
+    GLM_CSCOORD_LBF,
+    GLM_CSCOORD_LTF,
+    GLM_CSCOORD_RTF,
+    GLM_CSCOORD_RBF
+  };
+
+  glm_mat4_mulv(invMat, csCoords[0], c[0]);
+  glm_mat4_mulv(invMat, csCoords[1], c[1]);
+  glm_mat4_mulv(invMat, csCoords[2], c[2]);
+  glm_mat4_mulv(invMat, csCoords[3], c[3]);
+  glm_mat4_mulv(invMat, csCoords[4], c[4]);
+  glm_mat4_mulv(invMat, csCoords[5], c[5]);
+  glm_mat4_mulv(invMat, csCoords[6], c[6]);
+  glm_mat4_mulv(invMat, csCoords[7], c[7]);
+
+  glm_vec4_scale(c[0], 1.0f / c[0][3], dest[0]);
+  glm_vec4_scale(c[1], 1.0f / c[1][3], dest[1]);
+  glm_vec4_scale(c[2], 1.0f / c[2][3], dest[2]);
+  glm_vec4_scale(c[3], 1.0f / c[3][3], dest[3]);
+  glm_vec4_scale(c[4], 1.0f / c[4][3], dest[4]);
+  glm_vec4_scale(c[5], 1.0f / c[5][3], dest[5]);
+  glm_vec4_scale(c[6], 1.0f / c[6][3], dest[6]);
+  glm_vec4_scale(c[7], 1.0f / c[7][3], dest[7]);
+}
+
+/*!
+ * @brief finds center of view frustum
+ *
+ * @param[in]  corners view frustum corners
+ * @param[out] dest    view frustum center
+ */
+CGLM_INLINE
+void
+glm_frustum_center(vec4 corners[8], vec4 dest) {
+  vec4 center;
+
+  glm_vec4_copy(corners[0], center);
+
+  glm_vec4_add(corners[1], center, center);
+  glm_vec4_add(corners[2], center, center);
+  glm_vec4_add(corners[3], center, center);
+  glm_vec4_add(corners[4], center, center);
+  glm_vec4_add(corners[5], center, center);
+  glm_vec4_add(corners[6], center, center);
+  glm_vec4_add(corners[7], center, center);
+
+  glm_vec4_scale(center, 0.125f, dest);
+}
+
+/*!
+ * @brief finds bounding box of frustum relative to given matrix e.g. view mat
+ *
+ * @param[in]  corners view frustum corners
+ * @param[in]  m       matrix to convert existing conners
+ * @param[out] box     bounding box as array [min, max]
+ */
+CGLM_INLINE
+void
+glm_frustum_box(vec4 corners[8], mat4 m, vec3 box[2]) {
+  vec4 v;
+  vec3 min, max;
+  int  i;
+
+  glm_vec3_broadcast(FLT_MAX, min);
+  glm_vec3_broadcast(-FLT_MAX, max);
+
+  for (i = 0; i < 8; i++) {
+    glm_mat4_mulv(m, corners[i], v);
+
+    min[0] = glm_min(min[0], v[0]);
+    min[1] = glm_min(min[1], v[1]);
+    min[2] = glm_min(min[2], v[2]);
+
+    max[0] = glm_max(max[0], v[0]);
+    max[1] = glm_max(max[1], v[1]);
+    max[2] = glm_max(max[2], v[2]);
+  }
+
+  glm_vec3_copy(min, box[0]);
+  glm_vec3_copy(max, box[1]);
+}
+
+/*!
+ * @brief finds planes corners which is between near and far planes (parallel)
+ *
+ * this will be helpful if you want to split a frustum e.g. CSM/PSSM. This will
+ * find planes' corners but you will need to one more plane.
+ * Actually you have it, it is near, far or created previously with this func ;)
+ *
+ * @param[in]  corners view  frustum corners
+ * @param[in]  splitDist     split distance
+ * @param[in]  farDist       far distance (zFar)
+ * @param[out] planeCorners  plane corners [LB, LT, RT, RB]
+ */
+CGLM_INLINE
+void
+glm_frustum_corners_at(vec4  corners[8],
+                       float splitDist,
+                       float farDist,
+                       vec4  planeCorners[4]) {
+  vec4  corner;
+  float dist, sc;
+
+  /* because distance and scale is same for all */
+  dist = glm_vec3_distance(corners[GLM_RTF], corners[GLM_RTN]);
+  sc   = dist * (splitDist / farDist);
+
+  /* left bottom */
+  glm_vec4_sub(corners[GLM_LBF], corners[GLM_LBN], corner);
+  glm_vec4_scale_as(corner, sc, corner);
+  glm_vec4_add(corners[GLM_LBN], corner, planeCorners[0]);
+
+  /* left top */
+  glm_vec4_sub(corners[GLM_LTF], corners[GLM_LTN], corner);
+  glm_vec4_scale_as(corner, sc, corner);
+  glm_vec4_add(corners[GLM_LTN], corner, planeCorners[1]);
+
+  /* right top */
+  glm_vec4_sub(corners[GLM_RTF], corners[GLM_RTN], corner);
+  glm_vec4_scale_as(corner, sc, corner);
+  glm_vec4_add(corners[GLM_RTN], corner, planeCorners[2]);
+
+  /* right bottom */
+  glm_vec4_sub(corners[GLM_RBF], corners[GLM_RBN], corner);
+  glm_vec4_scale_as(corner, sc, corner);
+  glm_vec4_add(corners[GLM_RBN], corner, planeCorners[3]);
+}
+
+#endif /* cglm_frustum_h */

include/cglm/io.h

@@ -25,7 +25,7 @@
 
 CGLM_INLINE
 void
-glm_mat4_print(mat4   matrix,
+glm_mat4_print(mat4              matrix,
                FILE * __restrict ostream) {
   int i;
   int j;
@@ -55,7 +55,7 @@ glm_mat4_print(mat4   matrix,
 
 CGLM_INLINE
 void
-glm_mat3_print(mat3 matrix,
+glm_mat3_print(mat3              matrix,
                FILE * __restrict ostream) {
   int i;
   int j;
@@ -85,7 +85,37 @@ glm_mat3_print(mat3 matrix,
 
 CGLM_INLINE
 void
-glm_vec4_print(vec4 vec,
+glm_mat2_print(mat2              matrix,
+               FILE * __restrict ostream) {
+  int i;
+  int j;
+
+#define m 2
+#define n 2
+
+  fprintf(ostream, "Matrix (float%dx%d):\n", m, n);
+
+  for (i = 0; i < m; i++) {
+    fprintf(ostream, "\t|");
+    for (j = 0; j < n; j++) {
+      fprintf(ostream, "%0.4f", matrix[j][i]);;
+
+      if (j != n - 1)
+        fprintf(ostream, "\t");
+    }
+
+    fprintf(ostream, "|\n");
+  }
+
+  fprintf(ostream, "\n");
+
+#undef m
+#undef n
+}
+
+CGLM_INLINE
+void
+glm_vec4_print(vec4              vec,
                FILE * __restrict ostream) {
   int i;
 
@@ -107,7 +137,7 @@ glm_vec4_print(vec4 vec,
 
 CGLM_INLINE
 void
-glm_vec3_print(vec3 vec,
+glm_vec3_print(vec3              vec,
                FILE * __restrict ostream) {
   int i;
 
@@ -129,7 +159,7 @@ glm_vec3_print(vec3 vec,
 
 CGLM_INLINE
 void
-glm_ivec3_print(ivec3 vec,
+glm_ivec3_print(ivec3             vec,
                 FILE * __restrict ostream) {
   int i;
 
@@ -151,7 +181,29 @@ glm_ivec3_print(ivec3 vec,
 
 CGLM_INLINE
 void
-glm_versor_print(versor vec,
+glm_vec2_print(vec2              vec,
+               FILE * __restrict ostream) {
+  int i;
+
+#define m 2
+
+  fprintf(ostream, "Vector (float%d):\n\t|", m);
+
+  for (i = 0; i < m; i++) {
+    fprintf(ostream, "%0.4f", vec[i]);
+
+    if (i != m - 1)
+      fprintf(ostream, "\t");
+  }
+
+  fprintf(ostream, "|\n\n");
+
+#undef m
+}
+
+CGLM_INLINE
+void
+glm_versor_print(versor            vec,
                  FILE * __restrict ostream) {
   int i;
 
@@ -171,4 +223,33 @@ glm_versor_print(versor vec,
 #undef m
 }
 
+CGLM_INLINE
+void
+glm_aabb_print(vec3                    bbox[2],
+               const char * __restrict tag,
+               FILE       * __restrict ostream) {
+  int i, j;
+
+#define m 3
+
+  fprintf(ostream, "AABB (%s):\n", tag ? tag: "float");
+
+  for (i = 0; i < 2; i++) {
+    fprintf(ostream, "\t|");
+
+    for (j = 0; j < m; j++) {
+      fprintf(ostream, "%0.4f", bbox[i][j]);
+
+      if (j != m - 1)
+        fprintf(ostream, "\t");
+    }
+
+    fprintf(ostream, "|\n");
+  }
+
+  fprintf(ostream, "\n");
+
+#undef m
+}
+
 #endif /* cglm_io_h */

include/cglm/mat2.h

@@ -0,0 +1,322 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+/*
+ Macros:
+   GLM_MAT2_IDENTITY_INIT
+   GLM_MAT2_ZERO_INIT
+   GLM_MAT2_IDENTITY
+   GLM_MAT2_ZERO
+
+ Functions:
+   CGLM_INLINE void  glm_mat2_copy(mat2 mat, mat2 dest)
+   CGLM_INLINE void  glm_mat2_identity(mat2 mat)
+   CGLM_INLINE void  glm_mat2_identity_array(mat2 * restrict mat, size_t count)
+   CGLM_INLINE void  glm_mat2_zero(mat2 mat)
+   CGLM_INLINE void  glm_mat2_mul(mat2 m1, mat2 m2, mat2 dest)
+   CGLM_INLINE void  glm_mat2_transpose_to(mat2 m, mat2 dest)
+   CGLM_INLINE void  glm_mat2_transpose(mat2 m)
+   CGLM_INLINE void  glm_mat2_mulv(mat2 m, vec2 v, vec2 dest)
+   CGLM_INLINE float glm_mat2_trace(mat2 m)
+   CGLM_INLINE void  glm_mat2_scale(mat2 m, float s)
+   CGLM_INLINE float glm_mat2_det(mat2 mat)
+   CGLM_INLINE void  glm_mat2_inv(mat2 mat, mat2 dest)
+   CGLM_INLINE void  glm_mat2_swap_col(mat2 mat, int col1, int col2)
+   CGLM_INLINE void  glm_mat2_swap_row(mat2 mat, int row1, int row2)
+   CGLM_INLINE float glm_mat2_rmc(vec2 r, mat2 m, vec2 c)
+ */
+
+#ifndef cglm_mat2_h
+#define cglm_mat2_h
+
+#include "common.h"
+#include "vec2.h"
+
+#ifdef CGLM_SSE_FP
+#  include "simd/sse2/mat2.h"
+#endif
+
+#define GLM_MAT2_IDENTITY_INIT  {{1.0f, 0.0f}, {0.0f, 1.0f}}
+#define GLM_MAT2_ZERO_INIT      {{0.0f, 0.0f}, {0.0f, 0.0f}}
+
+/* for C only */
+#define GLM_MAT2_IDENTITY ((mat2)GLM_MAT2_IDENTITY_INIT)
+#define GLM_MAT2_ZERO     ((mat2)GLM_MAT2_ZERO_INIT)
+
+/*!
+ * @brief copy all members of [mat] to [dest]
+ *
+ * @param[in]  mat  source
+ * @param[out] dest destination
+ */
+CGLM_INLINE
+void
+glm_mat2_copy(mat2 mat, mat2 dest) {
+  glm_vec4_copy(mat[0], dest[0]);
+}
+
+/*!
+ * @brief make given matrix identity. It is identical with below,
+ *        but it is more easy to do that with this func especially for members
+ *        e.g. glm_mat2_identity(aStruct->aMatrix);
+ *
+ * @code
+ * glm_mat2_copy(GLM_MAT2_IDENTITY, mat); // C only
+ *
+ * // or
+ * mat2 mat = GLM_MAT2_IDENTITY_INIT;
+ * @endcode
+ *
+ * @param[in, out]  mat  destination
+ */
+CGLM_INLINE
+void
+glm_mat2_identity(mat2 mat) {
+  CGLM_ALIGN_MAT mat2 t = GLM_MAT2_IDENTITY_INIT;
+  glm_mat2_copy(t, mat);
+}
+
+/*!
+ * @brief make given matrix array's each element identity matrix
+ *
+ * @param[in, out]  mat   matrix array (must be aligned (16)
+ *                        if alignment is not disabled)
+ *
+ * @param[in]       count count of matrices
+ */
+CGLM_INLINE
+void
+glm_mat2_identity_array(mat2 * __restrict mat, size_t count) {
+  CGLM_ALIGN_MAT mat2 t = GLM_MAT2_IDENTITY_INIT;
+  size_t i;
+
+  for (i = 0; i < count; i++) {
+    glm_mat2_copy(t, mat[i]);
+  }
+}
+
+/*!
+ * @brief make given matrix zero.
+ *
+ * @param[in, out]  mat  matrix
+ */
+CGLM_INLINE
+void
+glm_mat2_zero(mat2 mat) {
+  CGLM_ALIGN_MAT mat2 t = GLM_MAT2_ZERO_INIT;
+  glm_mat2_copy(t, mat);
+}
+
+/*!
+ * @brief multiply m1 and m2 to dest
+ *
+ * m1, m2 and dest matrices can be same matrix, it is possible to write this:
+ *
+ * @code
+ * mat2 m = GLM_MAT2_IDENTITY_INIT;
+ * glm_mat2_mul(m, m, m);
+ * @endcode
+ *
+ * @param[in]  m1   left matrix
+ * @param[in]  m2   right matrix
+ * @param[out] dest destination matrix
+ */
+CGLM_INLINE
+void
+glm_mat2_mul(mat2 m1, mat2 m2, mat2 dest) {
+#if defined( __SSE__ ) || defined( __SSE2__ )
+  glm_mat2_mul_sse2(m1, m2, dest);
+#else
+  float a00 = m1[0][0], a01 = m1[0][1],
+        a10 = m1[1][0], a11 = m1[1][1],
+        b00 = m2[0][0], b01 = m2[0][1],
+        b10 = m2[1][0], b11 = m2[1][1];
+
+  dest[0][0] = a00 * b00 + a10 * b01;
+  dest[0][1] = a01 * b00 + a11 * b01;
+  dest[1][0] = a00 * b10 + a10 * b11;
+  dest[1][1] = a01 * b10 + a11 * b11;
+#endif
+}
+
+/*!
+ * @brief transpose mat2 and store in dest
+ *
+ * source matrix will not be transposed unless dest is m
+ *
+ * @param[in]  m     matrix
+ * @param[out] dest  result
+ */
+CGLM_INLINE
+void
+glm_mat2_transpose_to(mat2 m, mat2 dest) {
+#if defined( __SSE__ ) || defined( __SSE2__ )
+  glm_mat2_transp_sse2(m, dest);
+#else
+  dest[0][0] = m[0][0];
+  dest[0][1] = m[1][0];
+  dest[1][0] = m[0][1];
+  dest[1][1] = m[1][1];
+#endif
+}
+
+/*!
+ * @brief tranpose mat2 and store result in same matrix
+ *
+ * @param[in, out] m source and dest
+ */
+CGLM_INLINE
+void
+glm_mat2_transpose(mat2 m) {
+  float tmp;
+  tmp     = m[0][1];
+  m[0][1] = m[1][0];
+  m[1][0] = tmp;
+}
+
+/*!
+ * @brief multiply mat2 with vec2 (column vector) and store in dest vector
+ *
+ * @param[in]  m    mat2 (left)
+ * @param[in]  v    vec2 (right, column vector)
+ * @param[out] dest vec2 (result, column vector)
+ */
+CGLM_INLINE
+void
+glm_mat2_mulv(mat2 m, vec2 v, vec2 dest) {
+  dest[0] = m[0][0] * v[0] + m[1][0] * v[1];
+  dest[1] = m[0][1] * v[0] + m[1][1] * v[1];
+}
+
+/*!
+ * @brief trace of matrix
+ *
+ * sum of the elements on the main diagonal from upper left to the lower right
+ *
+ * @param[in]  m matrix
+ */
+CGLM_INLINE
+float
+glm_mat2_trace(mat2 m) {
+  return m[0][0] + m[1][1];
+}
+
+/*!
+ * @brief scale (multiply with scalar) matrix
+ *
+ * multiply matrix with scalar
+ *
+ * @param[in, out] m matrix
+ * @param[in]      s scalar
+ */
+CGLM_INLINE
+void
+glm_mat2_scale(mat2 m, float s) {
+  glm_vec4_scale(m[0], s, m[0]);
+}
+
+/*!
+ * @brief mat2 determinant
+ *
+ * @param[in] mat matrix
+ *
+ * @return determinant
+ */
+CGLM_INLINE
+float
+glm_mat2_det(mat2 mat) {
+  return mat[0][0] * mat[1][1] - mat[1][0] * mat[0][1];
+}
+
+/*!
+ * @brief inverse mat2 and store in dest
+ *
+ * @param[in]  mat  matrix
+ * @param[out] dest inverse matrix
+ */
+CGLM_INLINE
+void
+glm_mat2_inv(mat2 mat, mat2 dest) {
+  float det;
+  float a = mat[0][0], b = mat[0][1],
+        c = mat[1][0], d = mat[1][1];
+
+  det = 1.0f / (a * d - b * c);
+
+  dest[0][0] =  d * det;
+  dest[0][1] = -b * det;
+  dest[1][0] = -c * det;
+  dest[1][1] =  a * det;
+}
+
+/*!
+ * @brief swap two matrix columns
+ *
+ * @param[in,out] mat  matrix
+ * @param[in]     col1 col1
+ * @param[in]     col2 col2
+ */
+CGLM_INLINE
+void
+glm_mat2_swap_col(mat2 mat, int col1, int col2) {
+  float a, b;
+
+  a = mat[col1][0];
+  b = mat[col1][1];
+
+  mat[col1][0] = mat[col2][0];
+  mat[col1][1] = mat[col2][1];
+
+  mat[col2][0] = a;
+  mat[col2][1] = b;
+}
+
+/*!
+ * @brief swap two matrix rows
+ *
+ * @param[in,out] mat  matrix
+ * @param[in]     row1 row1
+ * @param[in]     row2 row2
+ */
+CGLM_INLINE
+void
+glm_mat2_swap_row(mat2 mat, int row1, int row2) {
+  float a, b;
+
+  a = mat[0][row1];
+  b = mat[1][row1];
+
+  mat[0][row1] = mat[0][row2];
+  mat[1][row1] = mat[1][row2];
+
+  mat[0][row2] = a;
+  mat[1][row2] = b;
+}
+
+/*!
+ * @brief helper for  R (row vector) * M (matrix) * C (column vector)
+ *
+ * rmc stands for Row * Matrix * Column
+ *
+ * the result is scalar because R * M = Matrix1x2 (row vector),
+ * then Matrix1x2 * Vec2 (column vector) = Matrix1x1 (Scalar)
+ *
+ * @param[in]  r   row vector or matrix1x2
+ * @param[in]  m   matrix2x2
+ * @param[in]  c   column vector or matrix2x1
+ *
+ * @return scalar value e.g. Matrix1x1
+ */
+CGLM_INLINE
+float
+glm_mat2_rmc(vec2 r, mat2 m, vec2 c) {
+  vec2 tmp;
+  glm_mat2_mulv(m, c, tmp);
+  return glm_vec2_dot(r, tmp);
+}
+
+#endif /* cglm_mat2_h */

include/cglm/mat3.h

@@ -16,21 +16,27 @@
  Functions:
    CGLM_INLINE void  glm_mat3_copy(mat3 mat, mat3 dest);
    CGLM_INLINE void  glm_mat3_identity(mat3 mat);
+   CGLM_INLINE void  glm_mat3_identity_array(mat3 * restrict mat, size_t count);
+   CGLM_INLINE void  glm_mat3_zero(mat3 mat);
    CGLM_INLINE void  glm_mat3_mul(mat3 m1, mat3 m2, mat3 dest);
    CGLM_INLINE void  glm_mat3_transpose_to(mat3 m, mat3 dest);
    CGLM_INLINE void  glm_mat3_transpose(mat3 m);
    CGLM_INLINE void  glm_mat3_mulv(mat3 m, vec3 v, vec3 dest);
+   CGLM_INLINE float glm_mat3_trace(mat3 m);
+   CGLM_INLINE void  glm_mat3_quat(mat3 m, versor dest);
    CGLM_INLINE void  glm_mat3_scale(mat3 m, float s);
    CGLM_INLINE float glm_mat3_det(mat3 mat);
    CGLM_INLINE void  glm_mat3_inv(mat3 mat, mat3 dest);
    CGLM_INLINE void  glm_mat3_swap_col(mat3 mat, int col1, int col2);
    CGLM_INLINE void  glm_mat3_swap_row(mat3 mat, int row1, int row2);
+   CGLM_INLINE float glm_mat3_rmc(vec3 r, mat3 m, vec3 c);
  */
 
 #ifndef cglm_mat3_h
 #define cglm_mat3_h
 
 #include "common.h"
+#include "vec3.h"
 
 #ifdef CGLM_SSE_FP
 #  include "simd/sse2/mat3.h"
@@ -45,8 +51,8 @@
 
 
 /* for C only */
-#define GLM_MAT3_IDENTITY (mat3)GLM_MAT3_IDENTITY_INIT
-#define GLM_MAT3_ZERO     (mat3)GLM_MAT3_ZERO_INIT
+#define GLM_MAT3_IDENTITY ((mat3)GLM_MAT3_IDENTITY_INIT)
+#define GLM_MAT3_ZERO     ((mat3)GLM_MAT3_ZERO_INIT)
 
 /* DEPRECATED! use _copy, _ucopy versions */
 #define glm_mat3_dup(mat, dest) glm_mat3_copy(mat, dest)
@@ -60,7 +66,17 @@
 CGLM_INLINE
 void
 glm_mat3_copy(mat3 mat, mat3 dest) {
-  glm__memcpy(float, dest, mat, sizeof(mat3));
+  dest[0][0] = mat[0][0];
+  dest[0][1] = mat[0][1];
+  dest[0][2] = mat[0][2];
+
+  dest[1][0] = mat[1][0];
+  dest[1][1] = mat[1][1];
+  dest[1][2] = mat[1][2];
+
+  dest[2][0] = mat[2][0];
+  dest[2][1] = mat[2][1];
+  dest[2][2] = mat[2][2];
 }
 
 /*!
@@ -80,7 +96,38 @@ glm_mat3_copy(mat3 mat, mat3 dest) {
 CGLM_INLINE
 void
 glm_mat3_identity(mat3 mat) {
-  mat3 t = GLM_MAT3_IDENTITY_INIT;
+  CGLM_ALIGN_MAT mat3 t = GLM_MAT3_IDENTITY_INIT;
+  glm_mat3_copy(t, mat);
+}
+
+/*!
+ * @brief make given matrix array's each element identity matrix
+ *
+ * @param[in, out]  mat   matrix array (must be aligned (16/32)
+ *                        if alignment is not disabled)
+ *
+ * @param[in]       count count of matrices
+ */
+CGLM_INLINE
+void
+glm_mat3_identity_array(mat3 * __restrict mat, size_t count) {
+  CGLM_ALIGN_MAT mat3 t = GLM_MAT3_IDENTITY_INIT;
+  size_t i;
+
+  for (i = 0; i < count; i++) {
+    glm_mat3_copy(t, mat[i]);
+  }
+}
+
+/*!
+ * @brief make given matrix zero.
+ *
+ * @param[in, out]  mat  matrix
+ */
+CGLM_INLINE
+void
+glm_mat3_zero(mat3 mat) {
+  CGLM_ALIGN_MAT mat3 t = GLM_MAT3_ZERO_INIT;
   glm_mat3_copy(t, mat);
 }
 
@@ -154,7 +201,7 @@ glm_mat3_transpose_to(mat3 m, mat3 dest) {
 CGLM_INLINE
 void
 glm_mat3_transpose(mat3 m) {
-  mat3 tmp;
+  CGLM_ALIGN_MAT mat3 tmp;
 
   tmp[0][1] = m[1][0];
   tmp[0][2] = m[2][0];
@@ -186,6 +233,68 @@ glm_mat3_mulv(mat3 m, vec3 v, vec3 dest) {
   dest[2] = m[0][2] * v[0] + m[1][2] * v[1] + m[2][2] * v[2];
 }
 
+/*!
+ * @brief trace of matrix
+ *
+ * sum of the elements on the main diagonal from upper left to the lower right
+ *
+ * @param[in]  m matrix
+ */
+CGLM_INLINE
+float
+glm_mat3_trace(mat3 m) {
+  return m[0][0] + m[1][1] + m[2][2];
+}
+
+/*!
+ * @brief convert mat3 to quaternion
+ *
+ * @param[in]  m    rotation matrix
+ * @param[out] dest destination quaternion
+ */
+CGLM_INLINE
+void
+glm_mat3_quat(mat3 m, versor dest) {
+  float trace, r, rinv;
+
+  /* it seems using like m12 instead of m[1][2] causes extra instructions */
+
+  trace = m[0][0] + m[1][1] + m[2][2];
+  if (trace >= 0.0f) {
+    r       = sqrtf(1.0f + trace);
+    rinv    = 0.5f / r;
+
+    dest[0] = rinv * (m[1][2] - m[2][1]);
+    dest[1] = rinv * (m[2][0] - m[0][2]);
+    dest[2] = rinv * (m[0][1] - m[1][0]);
+    dest[3] = r    * 0.5f;
+  } else if (m[0][0] >= m[1][1] && m[0][0] >= m[2][2]) {
+    r       = sqrtf(1.0f - m[1][1] - m[2][2] + m[0][0]);
+    rinv    = 0.5f / r;
+
+    dest[0] = r    * 0.5f;
+    dest[1] = rinv * (m[0][1] + m[1][0]);
+    dest[2] = rinv * (m[0][2] + m[2][0]);
+    dest[3] = rinv * (m[1][2] - m[2][1]);
+  } else if (m[1][1] >= m[2][2]) {
+    r       = sqrtf(1.0f - m[0][0] - m[2][2] + m[1][1]);
+    rinv    = 0.5f / r;
+
+    dest[0] = rinv * (m[0][1] + m[1][0]);
+    dest[1] = r    * 0.5f;
+    dest[2] = rinv * (m[1][2] + m[2][1]);
+    dest[3] = rinv * (m[2][0] - m[0][2]);
+  } else {
+    r       = sqrtf(1.0f - m[0][0] - m[1][1] + m[2][2]);
+    rinv    = 0.5f / r;
+
+    dest[0] = rinv * (m[0][2] + m[2][0]);
+    dest[1] = rinv * (m[1][2] + m[2][1]);
+    dest[2] = r    * 0.5f;
+    dest[3] = rinv * (m[0][1] - m[1][0]);
+  }
+}
+
 /*!
  * @brief scale (multiply with scalar) matrix
  *
@@ -259,9 +368,9 @@ CGLM_INLINE
 void
 glm_mat3_swap_col(mat3 mat, int col1, int col2) {
   vec3 tmp;
-  glm_vec_copy(mat[col1], tmp);
-  glm_vec_copy(mat[col2], mat[col1]);
-  glm_vec_copy(tmp, mat[col2]);
+  glm_vec3_copy(mat[col1], tmp);
+  glm_vec3_copy(mat[col2], mat[col1]);
+  glm_vec3_copy(tmp, mat[col2]);
 }
 
 /*!
@@ -288,4 +397,26 @@ glm_mat3_swap_row(mat3 mat, int row1, int row2) {
   mat[2][row2] = tmp[2];
 }
 
+/*!
+ * @brief helper for  R (row vector) * M (matrix) * C (column vector)
+ *
+ * rmc stands for Row * Matrix * Column
+ *
+ * the result is scalar because R * M = Matrix1x3 (row vector),
+ * then Matrix1x3 * Vec3 (column vector) = Matrix1x1 (Scalar)
+ *
+ * @param[in]  r   row vector or matrix1x3
+ * @param[in]  m   matrix3x3
+ * @param[in]  c   column vector or matrix3x1
+ *
+ * @return scalar value e.g. Matrix1x1
+ */
+CGLM_INLINE
+float
+glm_mat3_rmc(vec3 r, mat3 m, vec3 c) {
+  vec3 tmp;
+  glm_mat3_mulv(m, c, tmp);
+  return glm_vec3_dot(r, tmp);
+}
+
 #endif /* cglm_mat3_h */

include/cglm/mat4.h

@@ -16,13 +16,13 @@
    GLM_MAT4_ZERO_INIT
    GLM_MAT4_IDENTITY
    GLM_MAT4_ZERO
-   glm_mat4_udup(mat, dest)
-   glm_mat4_dup(mat, dest)
 
  Functions:
    CGLM_INLINE void  glm_mat4_ucopy(mat4 mat, mat4 dest);
    CGLM_INLINE void  glm_mat4_copy(mat4 mat, mat4 dest);
    CGLM_INLINE void  glm_mat4_identity(mat4 mat);
+   CGLM_INLINE void  glm_mat4_identity_array(mat4 * restrict mat, size_t count);
+   CGLM_INLINE void  glm_mat4_zero(mat4 mat);
    CGLM_INLINE void  glm_mat4_pick3(mat4 mat, mat3 dest);
    CGLM_INLINE void  glm_mat4_pick3t(mat4 mat, mat3 dest);
    CGLM_INLINE void  glm_mat4_ins3(mat3 mat, mat4 dest);
@@ -30,21 +30,27 @@
    CGLM_INLINE void  glm_mat4_mulN(mat4 *matrices[], int len, mat4 dest);
    CGLM_INLINE void  glm_mat4_mulv(mat4 m, vec4 v, vec4 dest);
    CGLM_INLINE void  glm_mat4_mulv3(mat4 m, vec3 v, vec3 dest);
+   CGLM_INLINE float glm_mat4_trace(mat4 m);
+   CGLM_INLINE float glm_mat4_trace3(mat4 m);
+   CGLM_INLINE void  glm_mat4_quat(mat4 m, versor dest) ;
    CGLM_INLINE void  glm_mat4_transpose_to(mat4 m, mat4 dest);
    CGLM_INLINE void  glm_mat4_transpose(mat4 m);
    CGLM_INLINE void  glm_mat4_scale_p(mat4 m, float s);
    CGLM_INLINE void  glm_mat4_scale(mat4 m, float s);
    CGLM_INLINE float glm_mat4_det(mat4 mat);
    CGLM_INLINE void  glm_mat4_inv(mat4 mat, mat4 dest);
-   CGLM_INLINE void  glm_mat4_inv_precise(mat4 mat, mat4 dest);
+   CGLM_INLINE void  glm_mat4_inv_fast(mat4 mat, mat4 dest);
    CGLM_INLINE void  glm_mat4_swap_col(mat4 mat, int col1, int col2);
    CGLM_INLINE void  glm_mat4_swap_row(mat4 mat, int row1, int row2);
+   CGLM_INLINE float glm_mat4_rmc(vec4 r, mat4 m, vec4 c);
  */
 
 #ifndef cglm_mat_h
 #define cglm_mat_h
 
 #include "common.h"
+#include "vec4.h"
+#include "vec3.h"
 
 #ifdef CGLM_SSE_FP
 #  include "simd/sse2/mat4.h"
@@ -58,7 +64,9 @@
 #  include "simd/neon/mat4.h"
 #endif
 
-#include <assert.h>
+#ifdef DEBUG
+# include <assert.h>
+#endif
 
 #define GLM_MAT4_IDENTITY_INIT  {{1.0f, 0.0f, 0.0f, 0.0f},                    \
                                  {0.0f, 1.0f, 0.0f, 0.0f},                    \
@@ -71,13 +79,16 @@
                                  {0.0f, 0.0f, 0.0f, 0.0f}}
 
 /* for C only */
-#define GLM_MAT4_IDENTITY (mat4)GLM_MAT4_IDENTITY_INIT
-#define GLM_MAT4_ZERO     (mat4)GLM_MAT4_ZERO_INIT
+#define GLM_MAT4_IDENTITY ((mat4)GLM_MAT4_IDENTITY_INIT)
+#define GLM_MAT4_ZERO     ((mat4)GLM_MAT4_ZERO_INIT)
 
 /* DEPRECATED! use _copy, _ucopy versions */
 #define glm_mat4_udup(mat, dest) glm_mat4_ucopy(mat, dest)
 #define glm_mat4_dup(mat, dest)  glm_mat4_copy(mat, dest)
 
+/* DEPRECATED! default is precise now. */
+#define glm_mat4_inv_precise(mat, dest) glm_mat4_inv(mat, dest)
+
 /*!
  * @brief copy all members of [mat] to [dest]
  *
@@ -90,7 +101,15 @@
 CGLM_INLINE
 void
 glm_mat4_ucopy(mat4 mat, mat4 dest) {
-  glm__memcpy(float, dest, mat, sizeof(mat4));
+  dest[0][0] = mat[0][0];  dest[1][0] = mat[1][0];
+  dest[0][1] = mat[0][1];  dest[1][1] = mat[1][1];
+  dest[0][2] = mat[0][2];  dest[1][2] = mat[1][2];
+  dest[0][3] = mat[0][3];  dest[1][3] = mat[1][3];
+
+  dest[2][0] = mat[2][0];  dest[3][0] = mat[3][0];
+  dest[2][1] = mat[2][1];  dest[3][1] = mat[3][1];
+  dest[2][2] = mat[2][2];  dest[3][2] = mat[3][2];
+  dest[2][3] = mat[2][3];  dest[3][3] = mat[3][3];
 }
 
 /*!
@@ -103,13 +122,18 @@ CGLM_INLINE
 void
 glm_mat4_copy(mat4 mat, mat4 dest) {
 #ifdef __AVX__
-  _mm256_store_ps(dest[0], _mm256_load_ps(mat[0]));
-  _mm256_store_ps(dest[2], _mm256_load_ps(mat[2]));
+  glmm_store256(dest[0], glmm_load256(mat[0]));
+  glmm_store256(dest[2], glmm_load256(mat[2]));
 #elif defined( __SSE__ ) || defined( __SSE2__ )
-  _mm_store_ps(dest[0], _mm_load_ps(mat[0]));
-  _mm_store_ps(dest[1], _mm_load_ps(mat[1]));
-  _mm_store_ps(dest[2], _mm_load_ps(mat[2]));
-  _mm_store_ps(dest[3], _mm_load_ps(mat[3]));
+  glmm_store(dest[0], glmm_load(mat[0]));
+  glmm_store(dest[1], glmm_load(mat[1]));
+  glmm_store(dest[2], glmm_load(mat[2]));
+  glmm_store(dest[3], glmm_load(mat[3]));
+#elif defined(CGLM_NEON_FP)
+  vst1q_f32(dest[0], vld1q_f32(mat[0]));
+  vst1q_f32(dest[1], vld1q_f32(mat[1]));
+  vst1q_f32(dest[2], vld1q_f32(mat[2]));
+  vst1q_f32(dest[3], vld1q_f32(mat[3]));
 #else
   glm_mat4_ucopy(mat, dest);
 #endif
@@ -132,7 +156,38 @@ glm_mat4_copy(mat4 mat, mat4 dest) {
 CGLM_INLINE
 void
 glm_mat4_identity(mat4 mat) {
-  mat4 t = GLM_MAT4_IDENTITY_INIT;
+  CGLM_ALIGN_MAT mat4 t = GLM_MAT4_IDENTITY_INIT;
+  glm_mat4_copy(t, mat);
+}
+
+/*!
+ * @brief make given matrix array's each element identity matrix
+ *
+ * @param[in, out]  mat   matrix array (must be aligned (16/32)
+ *                        if alignment is not disabled)
+ *
+ * @param[in]       count count of matrices
+ */
+CGLM_INLINE
+void
+glm_mat4_identity_array(mat4 * __restrict mat, size_t count) {
+  CGLM_ALIGN_MAT mat4 t = GLM_MAT4_IDENTITY_INIT;
+  size_t i;
+
+  for (i = 0; i < count; i++) {
+    glm_mat4_copy(t, mat[i]);
+  }
+}
+
+/*!
+ * @brief make given matrix zero.
+ *
+ * @param[in, out]  mat  matrix
+ */
+CGLM_INLINE
+void
+glm_mat4_zero(mat4 mat) {
+  CGLM_ALIGN_MAT mat4 t = GLM_MAT4_ZERO_INIT;
   glm_mat4_copy(t, mat);
 }
 
@@ -225,7 +280,7 @@ glm_mat4_mul(mat4 m1, mat4 m2, mat4 dest) {
   glm_mat4_mul_avx(m1, m2, dest);
 #elif defined( __SSE__ ) || defined( __SSE2__ )
   glm_mat4_mul_sse2(m1, m2, dest);
-#elif defined( __ARM_NEON_FP )
+#elif defined(CGLM_NEON_FP)
   glm_mat4_mul_neon(m1, m2, dest);
 #else
   float a00 = m1[0][0], a01 = m1[0][1], a02 = m1[0][2], a03 = m1[0][3],
@@ -278,19 +333,17 @@ glm_mat4_mul(mat4 m1, mat4 m2, mat4 dest) {
  */
 CGLM_INLINE
 void
-glm_mat4_mulN(mat4 * __restrict matrices[], int len, mat4 dest) {
-  int i;
+glm_mat4_mulN(mat4 * __restrict matrices[], uint32_t len, mat4 dest) {
+  uint32_t i;
 
+#ifdef DEBUG
   assert(len > 1 && "there must be least 2 matrices to go!");
+#endif
 
-  glm_mat4_mul(*matrices[0],
-               *matrices[1],
-               dest);
+  glm_mat4_mul(*matrices[0], *matrices[1], dest);
 
   for (i = 2; i < len; i++)
-    glm_mat4_mul(dest,
-                 *matrices[i],
-                 dest);
+    glm_mat4_mul(dest, *matrices[i], dest);
 }
 
 /*!
@@ -316,20 +369,98 @@ glm_mat4_mulv(mat4 m, vec4 v, vec4 dest) {
 }
 
 /*!
- * @brief multiply vector with mat4's mat3 part(rotation)
+ * @brief trace of matrix
  *
- * @param[in]  m    mat4(affine transform)
- * @param[in]  v    vec3
- * @param[out] dest vec3
+ * sum of the elements on the main diagonal from upper left to the lower right
+ *
+ * @param[in]  m matrix
+ */
+CGLM_INLINE
+float
+glm_mat4_trace(mat4 m) {
+  return m[0][0] + m[1][1] + m[2][2] + m[3][3];
+}
+
+/*!
+ * @brief trace of matrix (rotation part)
+ *
+ * sum of the elements on the main diagonal from upper left to the lower right
+ *
+ * @param[in]  m matrix
+ */
+CGLM_INLINE
+float
+glm_mat4_trace3(mat4 m) {
+  return m[0][0] + m[1][1] + m[2][2];
+}
+
+/*!
+ * @brief convert mat4's rotation part to quaternion
+ *
+ * @param[in]  m    affine matrix
+ * @param[out] dest destination quaternion
  */
 CGLM_INLINE
 void
-glm_mat4_mulv3(mat4 m, vec3 v, vec3 dest) {
-  vec3 res;
-  res[0] = m[0][0] * v[0] + m[1][0] * v[1] + m[2][0] * v[2];
-  res[1] = m[0][1] * v[0] + m[1][1] * v[1] + m[2][1] * v[2];
-  res[2] = m[0][2] * v[0] + m[1][2] * v[1] + m[2][2] * v[2];
-  glm_vec_copy(res, dest);
+glm_mat4_quat(mat4 m, versor dest) {
+  float trace, r, rinv;
+
+  /* it seems using like m12 instead of m[1][2] causes extra instructions */
+
+  trace = m[0][0] + m[1][1] + m[2][2];
+  if (trace >= 0.0f) {
+    r       = sqrtf(1.0f + trace);
+    rinv    = 0.5f / r;
+
+    dest[0] = rinv * (m[1][2] - m[2][1]);
+    dest[1] = rinv * (m[2][0] - m[0][2]);
+    dest[2] = rinv * (m[0][1] - m[1][0]);
+    dest[3] = r    * 0.5f;
+  } else if (m[0][0] >= m[1][1] && m[0][0] >= m[2][2]) {
+    r       = sqrtf(1.0f - m[1][1] - m[2][2] + m[0][0]);
+    rinv    = 0.5f / r;
+
+    dest[0] = r    * 0.5f;
+    dest[1] = rinv * (m[0][1] + m[1][0]);
+    dest[2] = rinv * (m[0][2] + m[2][0]);
+    dest[3] = rinv * (m[1][2] - m[2][1]);
+  } else if (m[1][1] >= m[2][2]) {
+    r       = sqrtf(1.0f - m[0][0] - m[2][2] + m[1][1]);
+    rinv    = 0.5f / r;
+
+    dest[0] = rinv * (m[0][1] + m[1][0]);
+    dest[1] = r    * 0.5f;
+    dest[2] = rinv * (m[1][2] + m[2][1]);
+    dest[3] = rinv * (m[2][0] - m[0][2]);
+  } else {
+    r       = sqrtf(1.0f - m[0][0] - m[1][1] + m[2][2]);
+    rinv    = 0.5f / r;
+
+    dest[0] = rinv * (m[0][2] + m[2][0]);
+    dest[1] = rinv * (m[1][2] + m[2][1]);
+    dest[2] = r    * 0.5f;
+    dest[3] = rinv * (m[0][1] - m[1][0]);
+  }
+}
+
+/*!
+ * @brief multiply vector with mat4
+ *
+ * actually the result is vec4, after multiplication the last component
+ * is trimmed. if you need it don't use this func.
+ *
+ * @param[in]  m    mat4(affine transform)
+ * @param[in]  v    vec3
+ * @param[in]  last 4th item to make it vec4
+ * @param[out] dest result vector (vec3)
+ */
+CGLM_INLINE
+void
+glm_mat4_mulv3(mat4 m, vec3 v, float last, vec3 dest) {
+  vec4 res;
+  glm_vec4(v, last, res);
+  glm_mat4_mulv(m, res, res);
+  glm_vec3(res, dest);
 }
 
 /*!
@@ -369,10 +500,8 @@ glm_mat4_transpose(mat4 m) {
   glm_mat4_transp_sse2(m, m);
 #else
   mat4 d;
-
   glm_mat4_transpose_to(m, d);
-
-  glm__memcpy(float, m, d, sizeof(mat4));
+  glm_mat4_ucopy(d, m);
 #endif
 }
 
@@ -406,6 +535,13 @@ void
 glm_mat4_scale(mat4 m, float s) {
 #if defined( __SSE__ ) || defined( __SSE2__ )
   glm_mat4_scale_sse2(m, s);
+#elif defined(CGLM_NEON_FP)
+  float32x4_t v0;
+  v0 = vdupq_n_f32(s);
+  vst1q_f32(m[0], vmulq_f32(vld1q_f32(m[0]), v0));
+  vst1q_f32(m[1], vmulq_f32(vld1q_f32(m[1]), v0));
+  vst1q_f32(m[2], vmulq_f32(vld1q_f32(m[2]), v0));
+  vst1q_f32(m[3], vmulq_f32(vld1q_f32(m[3]), v0));
 #else
   glm_mat4_scale_p(m, s);
 #endif
@@ -448,10 +584,6 @@ glm_mat4_det(mat4 mat) {
 /*!
  * @brief inverse mat4 and store in dest
  *
- * this func uses reciprocal approximation without extra corrections 
- * e.g Newton-Raphson. this should work faster than _precise, 
- * to get precise value use _precise version
- *
  * @param[in]  mat  matrix
  * @param[out] dest inverse matrix
  */
@@ -504,22 +636,23 @@ glm_mat4_inv(mat4 mat, mat4 dest) {
 #endif
 }
 
-
 /*!
- * @brief inverse mat4 precisely and store in dest
+ * @brief inverse mat4 and store in dest
  *
- * this do same thing as glm_mat4_inv did. the only diff is this func uses
- * division instead of reciprocal approximation. Due to division this might
- * work slower than glm_mat4_inv
+ * this func uses reciprocal approximation without extra corrections
+ * e.g Newton-Raphson. this should work faster than normal,
+ * to get more precise use glm_mat4_inv version.
+ *
+ * NOTE: You will lose precision, glm_mat4_inv is more accurate
  *
  * @param[in]  mat  matrix
  * @param[out] dest inverse matrix
  */
 CGLM_INLINE
 void
-glm_mat4_inv_precise(mat4 mat, mat4 dest) {
+glm_mat4_inv_fast(mat4 mat, mat4 dest) {
 #if defined( __SSE__ ) || defined( __SSE2__ )
-  glm_mat4_inv_precise_sse2(mat, dest);
+  glm_mat4_inv_fast_sse2(mat, dest);
 #else
   glm_mat4_inv(mat, dest);
 #endif
@@ -535,7 +668,7 @@ glm_mat4_inv_precise(mat4 mat, mat4 dest) {
 CGLM_INLINE
 void
 glm_mat4_swap_col(mat4 mat, int col1, int col2) {
-  vec4 tmp;
+  CGLM_ALIGN(16) vec4 tmp;
   glm_vec4_copy(mat[col1], tmp);
   glm_vec4_copy(mat[col2], mat[col1]);
   glm_vec4_copy(tmp, mat[col2]);
@@ -551,7 +684,7 @@ glm_mat4_swap_col(mat4 mat, int col1, int col2) {
 CGLM_INLINE
 void
 glm_mat4_swap_row(mat4 mat, int row1, int row2) {
-  vec4 tmp;
+  CGLM_ALIGN(16) vec4 tmp;
   tmp[0] = mat[0][row1];
   tmp[1] = mat[1][row1];
   tmp[2] = mat[2][row1];
@@ -568,5 +701,26 @@ glm_mat4_swap_row(mat4 mat, int row1, int row2) {
   mat[3][row2] = tmp[3];
 }
 
-#else
+/*!
+ * @brief helper for  R (row vector) * M (matrix) * C (column vector)
+ *
+ * rmc stands for Row * Matrix * Column
+ *
+ * the result is scalar because R * M = Matrix1x4 (row vector),
+ * then Matrix1x4 * Vec4 (column vector) = Matrix1x1 (Scalar)
+ *
+ * @param[in]  r   row vector or matrix1x4
+ * @param[in]  m   matrix4x4
+ * @param[in]  c   column vector or matrix4x1
+ *
+ * @return scalar value e.g. B(s)
+ */
+CGLM_INLINE
+float
+glm_mat4_rmc(vec4 r, mat4 m, vec4 c) {
+  vec4 tmp;
+  glm_mat4_mulv(m, c, tmp);
+  return glm_vec4_dot(r, tmp);
+}
+
 #endif /* cglm_mat_h */

include/cglm/plane.h

@@ -0,0 +1,43 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglm_plane_h
+#define cglm_plane_h
+
+#include "common.h"
+#include "vec4.h"
+
+/*
+ Plane equation:  Ax + By + Cz + D = 0;
+
+ It stored in vec4 as [A, B, C, D]. (A, B, C) is normal and D is distance
+*/
+
+/*
+ Functions:
+   CGLM_INLINE void  glm_plane_normalize(vec4 plane);
+ */
+
+/*!
+ * @brief normalizes a plane
+ *
+ * @param[in, out] plane plane to normalize
+ */
+CGLM_INLINE
+void
+glm_plane_normalize(vec4 plane) {
+  float norm;
+  
+  if ((norm = glm_vec3_norm(plane)) == 0.0f) {
+    glm_vec4_zero(plane);
+    return;
+  }
+  
+  glm_vec4_scale(plane, 1.0f / norm, plane);
+}
+
+#endif /* cglm_plane_h */

include/cglm/project.h

@@ -0,0 +1,118 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglm_project_h
+#define cglm_project_h
+
+#include "common.h"
+#include "vec3.h"
+#include "vec4.h"
+#include "mat4.h"
+
+/*!
+ * @brief maps the specified viewport coordinates into specified space [1]
+ *        the matrix should contain projection matrix.
+ *
+ * if you don't have ( and don't want to have ) an inverse matrix then use
+ * glm_unproject version. You may use existing inverse of matrix in somewhere
+ * else, this is why glm_unprojecti exists to save save inversion cost
+ *
+ * [1] space:
+ *  1- if m = invProj:     View Space
+ *  2- if m = invViewProj: World Space
+ *  3- if m = invMVP:      Object Space
+ *
+ * You probably want to map the coordinates into object space
+ * so use invMVP as m
+ *
+ * Computing viewProj:
+ *   glm_mat4_mul(proj, view, viewProj);
+ *   glm_mat4_mul(viewProj, model, MVP);
+ *   glm_mat4_inv(viewProj, invMVP);
+ *
+ * @param[in]  pos      point/position in viewport coordinates
+ * @param[in]  invMat   matrix (see brief)
+ * @param[in]  vp       viewport as [x, y, width, height]
+ * @param[out] dest     unprojected coordinates
+ */
+CGLM_INLINE
+void
+glm_unprojecti(vec3 pos, mat4 invMat, vec4 vp, vec3 dest) {
+  vec4 v;
+
+  v[0] = 2.0f * (pos[0] - vp[0]) / vp[2] - 1.0f;
+  v[1] = 2.0f * (pos[1] - vp[1]) / vp[3] - 1.0f;
+  v[2] = 2.0f *  pos[2]                  - 1.0f;
+  v[3] = 1.0f;
+
+  glm_mat4_mulv(invMat, v, v);
+  glm_vec4_scale(v, 1.0f / v[3], v);
+  glm_vec3(v, dest);
+}
+
+/*!
+ * @brief maps the specified viewport coordinates into specified space [1]
+ *        the matrix should contain projection matrix.
+ *
+ * this is same as glm_unprojecti except this function get inverse matrix for
+ * you.
+ *
+ * [1] space:
+ *  1- if m = proj:     View Space
+ *  2- if m = viewProj: World Space
+ *  3- if m = MVP:      Object Space
+ *
+ * You probably want to map the coordinates into object space
+ * so use MVP as m
+ *
+ * Computing viewProj and MVP:
+ *   glm_mat4_mul(proj, view, viewProj);
+ *   glm_mat4_mul(viewProj, model, MVP);
+ *
+ * @param[in]  pos      point/position in viewport coordinates
+ * @param[in]  m        matrix (see brief)
+ * @param[in]  vp       viewport as [x, y, width, height]
+ * @param[out] dest     unprojected coordinates
+ */
+CGLM_INLINE
+void
+glm_unproject(vec3 pos, mat4 m, vec4 vp, vec3 dest) {
+  mat4 inv;
+  glm_mat4_inv(m, inv);
+  glm_unprojecti(pos, inv, vp, dest);
+}
+
+/*!
+ * @brief map object coordinates to window coordinates
+ *
+ * Computing MVP:
+ *   glm_mat4_mul(proj, view, viewProj);
+ *   glm_mat4_mul(viewProj, model, MVP);
+ *
+ * @param[in]  pos      object coordinates
+ * @param[in]  m        MVP matrix
+ * @param[in]  vp       viewport as [x, y, width, height]
+ * @param[out] dest     projected coordinates
+ */
+CGLM_INLINE
+void
+glm_project(vec3 pos, mat4 m, vec4 vp, vec3 dest) {
+  CGLM_ALIGN(16) vec4 pos4, vone = GLM_VEC4_ONE_INIT;
+
+  glm_vec4(pos, 1.0f, pos4);
+
+  glm_mat4_mulv(m, pos4, pos4);
+  glm_vec4_scale(pos4, 1.0f / pos4[3], pos4); /* pos = pos / pos.w */
+  glm_vec4_add(pos4, vone, pos4);
+  glm_vec4_scale(pos4, 0.5f, pos4);
+
+  dest[0] = pos4[0] * vp[2] + vp[0];
+  dest[1] = pos4[1] * vp[3] + vp[1];
+  dest[2] = pos4[2];
+}
+
+#endif /* cglm_project_h */

include/cglm/quat.h

@@ -11,126 +11,444 @@
    GLM_QUAT_IDENTITY
 
  Functions:
-   CGLM_INLINE void  glm_quat_identity(versor q);
-   CGLM_INLINE void  glm_quat(versor q, float angle, float x, float y, float z);
-   CGLM_INLINE void  glm_quatv(versor q, float angle, vec3 v);
+   CGLM_INLINE void glm_quat_identity(versor q);
+   CGLM_INLINE void glm_quat_init(versor q, float x, float y, float z, float w);
+   CGLM_INLINE void glm_quat(versor q, float angle, float x, float y, float z);
+   CGLM_INLINE void glm_quatv(versor q, float angle, vec3 axis);
+   CGLM_INLINE void glm_quat_copy(versor q, versor dest);
    CGLM_INLINE float glm_quat_norm(versor q);
-   CGLM_INLINE void  glm_quat_normalize(versor q);
-   CGLM_INLINE float glm_quat_dot(versor q, versor r);
-   CGLM_INLINE void  glm_quat_mulv(versor q1, versor q2, versor dest);
-   CGLM_INLINE void  glm_quat_mat4(versor q, mat4 dest);
-   CGLM_INLINE void  glm_quat_slerp(versor q, versor r, float t, versor dest);
+   CGLM_INLINE void glm_quat_normalize(versor q);
+   CGLM_INLINE void glm_quat_normalize_to(versor q, versor dest);
+   CGLM_INLINE float glm_quat_dot(versor p, versor q);
+   CGLM_INLINE void glm_quat_conjugate(versor q, versor dest);
+   CGLM_INLINE void glm_quat_inv(versor q, versor dest);
+   CGLM_INLINE void glm_quat_add(versor p, versor q, versor dest);
+   CGLM_INLINE void glm_quat_sub(versor p, versor q, versor dest);
+   CGLM_INLINE float glm_quat_real(versor q);
+   CGLM_INLINE void glm_quat_imag(versor q, vec3 dest);
+   CGLM_INLINE void glm_quat_imagn(versor q, vec3 dest);
+   CGLM_INLINE float glm_quat_imaglen(versor q);
+   CGLM_INLINE float glm_quat_angle(versor q);
+   CGLM_INLINE void glm_quat_axis(versor q, vec3 dest);
+   CGLM_INLINE void glm_quat_mul(versor p, versor q, versor dest);
+   CGLM_INLINE void glm_quat_mat4(versor q, mat4 dest);
+   CGLM_INLINE void glm_quat_mat4t(versor q, mat4 dest);
+   CGLM_INLINE void glm_quat_mat3(versor q, mat3 dest);
+   CGLM_INLINE void glm_quat_mat3t(versor q, mat3 dest);
+   CGLM_INLINE void glm_quat_lerp(versor from, versor to, float t, versor dest);
+   CGLM_INLINE void glm_quat_lerpc(versor from, versor to, float t, versor dest);
+   CGLM_INLINE void glm_quat_slerp(versor q, versor r, float t, versor dest);
+   CGLM_INLINE void glm_quat_look(vec3 eye, versor ori, mat4 dest);
+   CGLM_INLINE void glm_quat_for(vec3 dir, vec3 fwd, vec3 up, versor dest);
+   CGLM_INLINE void glm_quat_forp(vec3 from,
+                                  vec3 to,
+                                  vec3 fwd,
+                                  vec3 up,
+                                  versor dest);
+   CGLM_INLINE void glm_quat_rotatev(versor q, vec3 v, vec3 dest);
+   CGLM_INLINE void glm_quat_rotate(mat4 m, versor q, mat4 dest);
  */
 
 #ifndef cglm_quat_h
 #define cglm_quat_h
 
 #include "common.h"
+#include "vec3.h"
 #include "vec4.h"
+#include "mat4.h"
+#include "mat3.h"
+#include "affine-mat.h"
 
 #ifdef CGLM_SSE_FP
 #  include "simd/sse2/quat.h"
 #endif
 
-#define GLM_QUAT_IDENTITY_INIT  {0.0f, 0.0f, 0.0f, 1.0f}
-#define GLM_QUAT_IDENTITY       (versor){0.0f, 0.0f, 0.0f, 1.0f}
+CGLM_INLINE
+void
+glm_mat4_identity(mat4 mat);
 
+CGLM_INLINE
+void
+glm_mat4_mulv(mat4 m, vec4 v, vec4 dest);
+
+CGLM_INLINE
+void
+glm_mul_rot(mat4 m1, mat4 m2, mat4 dest);
+
+CGLM_INLINE
+void
+glm_translate(mat4 m, vec3 v);
+
+/*
+ * IMPORTANT:
+ * ----------------------------------------------------------------------------
+ * cglm stores quat as [x, y, z, w] since v0.3.6
+ *
+ * it was [w, x, y, z] before v0.3.6 it has been changed to [x, y, z, w]
+ * with v0.3.6 version.
+ * ----------------------------------------------------------------------------
+ */
+
+#define GLM_QUAT_IDENTITY_INIT  {0.0f, 0.0f, 0.0f, 1.0f}
+#define GLM_QUAT_IDENTITY       ((versor)GLM_QUAT_IDENTITY_INIT)
+
+/*!
+ * @brief makes given quat to identity
+ *
+ * @param[in, out]  q  quaternion
+ */
 CGLM_INLINE
 void
 glm_quat_identity(versor q) {
-  versor v = GLM_QUAT_IDENTITY_INIT;
+  CGLM_ALIGN(16) versor v = GLM_QUAT_IDENTITY_INIT;
   glm_vec4_copy(v, q);
 }
 
+/*!
+ * @brief make given quaternion array's each element identity quaternion
+ *
+ * @param[in, out]  q     quat array (must be aligned (16)
+ *                        if alignment is not disabled)
+ *
+ * @param[in]       count count of quaternions
+ */
 CGLM_INLINE
 void
-glm_quat(versor q,
-         float  angle,
-         float  x,
-         float  y,
-         float  z) {
+glm_quat_identity_array(versor * __restrict q, size_t count) {
+  CGLM_ALIGN(16) versor v = GLM_QUAT_IDENTITY_INIT;
+  size_t i;
+
+  for (i = 0; i < count; i++) {
+    glm_vec4_copy(v, q[i]);
+  }
+}
+
+/*!
+ * @brief inits quaterion with raw values
+ *
+ * @param[out]  q     quaternion
+ * @param[in]   x     x
+ * @param[in]   y     y
+ * @param[in]   z     z
+ * @param[in]   w     w (real part)
+ */
+CGLM_INLINE
+void
+glm_quat_init(versor q, float x, float y, float z, float w) {
+  q[0] = x;
+  q[1] = y;
+  q[2] = z;
+  q[3] = w;
+}
+
+/*!
+ * @brief creates NEW quaternion with axis vector
+ *
+ * @param[out]  q     quaternion
+ * @param[in]   angle angle (radians)
+ * @param[in]   axis  axis
+ */
+CGLM_INLINE
+void
+glm_quatv(versor q, float angle, vec3 axis) {
+  CGLM_ALIGN(8) vec3 k;
   float a, c, s;
 
   a = angle * 0.5f;
   c = cosf(a);
   s = sinf(a);
 
-  q[0] = c;
-  q[1] = s * x;
-  q[2] = s * y;
-  q[3] = s * z;
+  glm_normalize_to(axis, k);
+
+  q[0] = s * k[0];
+  q[1] = s * k[1];
+  q[2] = s * k[2];
+  q[3] = c;
 }
 
+/*!
+ * @brief creates NEW quaternion with individual axis components
+ *
+ * @param[out]  q     quaternion
+ * @param[in]   angle angle (radians)
+ * @param[in]   x     axis.x
+ * @param[in]   y     axis.y
+ * @param[in]   z     axis.z
+ */
 CGLM_INLINE
 void
-glm_quatv(versor q,
-          float  angle,
-          vec3   v) {
-  float a, c, s;
-
-  a = angle * 0.5f;
-  c = cosf(a);
-  s = sinf(a);
-
-  q[0] = c;
-  q[1] = s * v[0];
-  q[2] = s * v[1];
-  q[3] = s * v[2];
+glm_quat(versor q, float angle, float x, float y, float z) {
+  CGLM_ALIGN(8) vec3 axis = {x, y, z};
+  glm_quatv(q, angle, axis);
 }
 
+/*!
+ * @brief copy quaternion to another one
+ *
+ * @param[in]  q     quaternion
+ * @param[out] dest  destination
+ */
+CGLM_INLINE
+void
+glm_quat_copy(versor q, versor dest) {
+  glm_vec4_copy(q, dest);
+}
+
+/*!
+ * @brief returns norm (magnitude) of quaternion
+ *
+ * @param[out]  q  quaternion
+ */
 CGLM_INLINE
 float
 glm_quat_norm(versor q) {
   return glm_vec4_norm(q);
 }
 
+/*!
+ * @brief normalize quaternion and store result in dest
+ *
+ * @param[in]   q     quaternion to normalze
+ * @param[out]  dest  destination quaternion
+ */
+CGLM_INLINE
+void
+glm_quat_normalize_to(versor q, versor dest) {
+#if defined( __SSE2__ ) || defined( __SSE2__ )
+  __m128 xdot, x0;
+  float  dot;
+
+  x0   = glmm_load(q);
+  xdot = glmm_vdot(x0, x0);
+  dot  = _mm_cvtss_f32(xdot);
+
+  if (dot <= 0.0f) {
+    glm_quat_identity(dest);
+    return;
+  }
+
+  glmm_store(dest, _mm_div_ps(x0, _mm_sqrt_ps(xdot)));
+#else
+  float dot;
+
+  dot = glm_vec4_norm2(q);
+
+  if (dot <= 0.0f) {
+    glm_quat_identity(dest);
+    return;
+  }
+
+  glm_vec4_scale(q, 1.0f / sqrtf(dot), dest);
+#endif
+}
+
+/*!
+ * @brief normalize quaternion
+ *
+ * @param[in, out]  q  quaternion
+ */
 CGLM_INLINE
 void
 glm_quat_normalize(versor q) {
-  float sum;
-
-  sum = q[0] * q[0] + q[1] * q[1]
-          + q[2] * q[2] + q[3] * q[3];
-
-  if (fabs(1.0f - sum) < 0.0001f)
-    return;
-
-  glm_vec4_scale(q, 1.0f / sqrtf(sum), q);
+  glm_quat_normalize_to(q, q);
 }
 
+/*!
+ * @brief dot product of two quaternion
+ *
+ * @param[in]  p  quaternion 1
+ * @param[in]  q  quaternion 2
+ */
 CGLM_INLINE
 float
-glm_quat_dot(versor q, versor r) {
-  return glm_vec4_dot(q, r);
+glm_quat_dot(versor p, versor q) {
+  return glm_vec4_dot(p, q);
 }
 
+/*!
+ * @brief conjugate of quaternion
+ *
+ * @param[in]   q     quaternion
+ * @param[out]  dest  conjugate
+ */
 CGLM_INLINE
 void
-glm_quat_mulv(versor q1, versor q2, versor dest) {
-  dest[0] = q2[0] * q1[0] - q2[1] * q1[1] - q2[2] * q1[2] - q2[3] * q1[3];
-  dest[1] = q2[0] * q1[1] + q2[1] * q1[0] - q2[2] * q1[3] + q2[3] * q1[2];
-  dest[2] = q2[0] * q1[2] + q2[1] * q1[3] + q2[2] * q1[0] - q2[3] * q1[1];
-  dest[3] = q2[0] * q1[3] - q2[1] * q1[2] + q2[2] * q1[1] + q2[3] * q1[0];
-
-  glm_quat_normalize(dest);
+glm_quat_conjugate(versor q, versor dest) {
+  glm_vec4_negate_to(q, dest);
+  dest[3] = -dest[3];
 }
 
+/*!
+ * @brief inverse of non-zero quaternion
+ *
+ * @param[in]   q    quaternion
+ * @param[out]  dest inverse quaternion
+ */
+CGLM_INLINE
+void
+glm_quat_inv(versor q, versor dest) {
+  CGLM_ALIGN(16) versor conj;
+  glm_quat_conjugate(q, conj);
+  glm_vec4_scale(conj, 1.0f / glm_vec4_norm2(q), dest);
+}
+
+/*!
+ * @brief add (componentwise) two quaternions and store result in dest
+ *
+ * @param[in]   p    quaternion 1
+ * @param[in]   q    quaternion 2
+ * @param[out]  dest result quaternion
+ */
+CGLM_INLINE
+void
+glm_quat_add(versor p, versor q, versor dest) {
+  glm_vec4_add(p, q, dest);
+}
+
+/*!
+ * @brief subtract (componentwise) two quaternions and store result in dest
+ *
+ * @param[in]   p    quaternion 1
+ * @param[in]   q    quaternion 2
+ * @param[out]  dest result quaternion
+ */
+CGLM_INLINE
+void
+glm_quat_sub(versor p, versor q, versor dest) {
+  glm_vec4_sub(p, q, dest);
+}
+
+/*!
+ * @brief returns real part of quaternion
+ *
+ * @param[in]   q    quaternion
+ */
+CGLM_INLINE
+float
+glm_quat_real(versor q) {
+  return q[3];
+}
+
+/*!
+ * @brief returns imaginary part of quaternion
+ *
+ * @param[in]   q    quaternion
+ * @param[out]  dest imag
+ */
+CGLM_INLINE
+void
+glm_quat_imag(versor q, vec3 dest) {
+  dest[0] = q[0];
+  dest[1] = q[1];
+  dest[2] = q[2];
+}
+
+/*!
+ * @brief returns normalized imaginary part of quaternion
+ *
+ * @param[in]   q    quaternion
+ */
+CGLM_INLINE
+void
+glm_quat_imagn(versor q, vec3 dest) {
+  glm_normalize_to(q, dest);
+}
+
+/*!
+ * @brief returns length of imaginary part of quaternion
+ *
+ * @param[in]   q    quaternion
+ */
+CGLM_INLINE
+float
+glm_quat_imaglen(versor q) {
+  return glm_vec3_norm(q);
+}
+
+/*!
+ * @brief returns angle of quaternion
+ *
+ * @param[in]   q    quaternion
+ */
+CGLM_INLINE
+float
+glm_quat_angle(versor q) {
+  /*
+   sin(theta / 2) = length(x*x + y*y + z*z)
+   cos(theta / 2) = w
+   theta          = 2 * atan(sin(theta / 2) / cos(theta / 2))
+   */
+  return 2.0f * atan2f(glm_quat_imaglen(q), glm_quat_real(q));
+}
+
+/*!
+ * @brief axis of quaternion
+ *
+ * @param[in]   q    quaternion
+ * @param[out]  dest axis of quaternion
+ */
+CGLM_INLINE
+void
+glm_quat_axis(versor q, vec3 dest) {
+  glm_quat_imagn(q, dest);
+}
+
+/*!
+ * @brief multiplies two quaternion and stores result in dest
+ *        this is also called Hamilton Product
+ *
+ * According to WikiPedia:
+ * The product of two rotation quaternions [clarification needed] will be
+ * equivalent to the rotation q followed by the rotation p
+ *
+ * @param[in]   p     quaternion 1
+ * @param[in]   q     quaternion 2
+ * @param[out]  dest  result quaternion
+ */
+CGLM_INLINE
+void
+glm_quat_mul(versor p, versor q, versor dest) {
+  /*
+    + (a1 b2 + b1 a2 + c1 d2 − d1 c2)i
+    + (a1 c2 − b1 d2 + c1 a2 + d1 b2)j
+    + (a1 d2 + b1 c2 − c1 b2 + d1 a2)k
+       a1 a2 − b1 b2 − c1 c2 − d1 d2
+   */
+#if defined( __SSE__ ) || defined( __SSE2__ )
+  glm_quat_mul_sse2(p, q, dest);
+#else
+  dest[0] = p[3] * q[0] + p[0] * q[3] + p[1] * q[2] - p[2] * q[1];
+  dest[1] = p[3] * q[1] - p[0] * q[2] + p[1] * q[3] + p[2] * q[0];
+  dest[2] = p[3] * q[2] + p[0] * q[1] - p[1] * q[0] + p[2] * q[3];
+  dest[3] = p[3] * q[3] - p[0] * q[0] - p[1] * q[1] - p[2] * q[2];
+#endif
+}
+
+/*!
+ * @brief convert quaternion to mat4
+ *
+ * @param[in]   q     quaternion
+ * @param[out]  dest  result matrix
+ */
 CGLM_INLINE
 void
 glm_quat_mat4(versor q, mat4 dest) {
-  float w, x, y, z;
-  float xx, yy, zz;
-  float xy, yz, xz;
-  float wx, wy, wz;
+  float w, x, y, z,
+        xx, yy, zz,
+        xy, yz, xz,
+        wx, wy, wz, norm, s;
 
-  w = q[0];
-  x = q[1];
-  y = q[2];
-  z = q[3];
+  norm = glm_quat_norm(q);
+  s    = norm > 0.0f ? 2.0f / norm : 0.0f;
 
-  xx = 2.0f * x * x;   xy = 2.0f * x * y;   wx = 2.0f * w * x;
-  yy = 2.0f * y * y;   yz = 2.0f * y * z;   wy = 2.0f * w * y;
-  zz = 2.0f * z * z;   xz = 2.0f * x * z;   wz = 2.0f * w * z;
+  x = q[0];
+  y = q[1];
+  z = q[2];
+  w = q[3];
+
+  xx = s * x * x;   xy = s * x * y;   wx = s * w * x;
+  yy = s * y * y;   yz = s * y * z;   wy = s * w * y;
+  zz = s * z * z;   xz = s * x * z;   wz = s * w * z;
 
   dest[0][0] = 1.0f - yy - zz;
   dest[1][1] = 1.0f - xx - zz;
@@ -144,8 +462,8 @@ glm_quat_mat4(versor q, mat4 dest) {
   dest[2][1] = yz - wx;
   dest[0][2] = xz - wy;
 
-  dest[1][3] = 0.0f;
   dest[0][3] = 0.0f;
+  dest[1][3] = 0.0f;
   dest[2][3] = 0.0f;
   dest[3][0] = 0.0f;
   dest[3][1] = 0.0f;
@@ -153,60 +471,348 @@ glm_quat_mat4(versor q, mat4 dest) {
   dest[3][3] = 1.0f;
 }
 
+/*!
+ * @brief convert quaternion to mat4 (transposed)
+ *
+ * @param[in]   q     quaternion
+ * @param[out]  dest  result matrix as transposed
+ */
 CGLM_INLINE
 void
-glm_quat_slerp(versor q,
-               versor r,
-               float  t,
-               versor dest) {
-  /* https://en.wikipedia.org/wiki/Slerp */
-#if defined( __SSE__ ) || defined( __SSE2__ )
-  glm_quat_slerp_sse2(q, r, t, dest);
-#else
-  float cosTheta, sinTheta, angle, a, b, c;
+glm_quat_mat4t(versor q, mat4 dest) {
+  float w, x, y, z,
+        xx, yy, zz,
+        xy, yz, xz,
+        wx, wy, wz, norm, s;
 
-  cosTheta = glm_quat_dot(q, r);
-  if (cosTheta < 0.0f) {
-    q[0] *= -1.0f;
-    q[1] *= -1.0f;
-    q[2] *= -1.0f;
-    q[3] *= -1.0f;
+  norm = glm_quat_norm(q);
+  s    = norm > 0.0f ? 2.0f / norm : 0.0f;
 
-    cosTheta = -cosTheta;
-  }
+  x = q[0];
+  y = q[1];
+  z = q[2];
+  w = q[3];
 
-  if (fabs(cosTheta) >= 1.0f) {
-    dest[0] = q[0];
-    dest[1] = q[1];
-    dest[2] = q[2];
-    dest[3] = q[3];
+  xx = s * x * x;   xy = s * x * y;   wx = s * w * x;
+  yy = s * y * y;   yz = s * y * z;   wy = s * w * y;
+  zz = s * z * z;   xz = s * x * z;   wz = s * w * z;
+
+  dest[0][0] = 1.0f - yy - zz;
+  dest[1][1] = 1.0f - xx - zz;
+  dest[2][2] = 1.0f - xx - yy;
+
+  dest[1][0] = xy + wz;
+  dest[2][1] = yz + wx;
+  dest[0][2] = xz + wy;
+
+  dest[0][1] = xy - wz;
+  dest[1][2] = yz - wx;
+  dest[2][0] = xz - wy;
+
+  dest[0][3] = 0.0f;
+  dest[1][3] = 0.0f;
+  dest[2][3] = 0.0f;
+  dest[3][0] = 0.0f;
+  dest[3][1] = 0.0f;
+  dest[3][2] = 0.0f;
+  dest[3][3] = 1.0f;
+}
+
+/*!
+ * @brief convert quaternion to mat3
+ *
+ * @param[in]   q     quaternion
+ * @param[out]  dest  result matrix
+ */
+CGLM_INLINE
+void
+glm_quat_mat3(versor q, mat3 dest) {
+  float w, x, y, z,
+        xx, yy, zz,
+        xy, yz, xz,
+        wx, wy, wz, norm, s;
+
+  norm = glm_quat_norm(q);
+  s    = norm > 0.0f ? 2.0f / norm : 0.0f;
+
+  x = q[0];
+  y = q[1];
+  z = q[2];
+  w = q[3];
+
+  xx = s * x * x;   xy = s * x * y;   wx = s * w * x;
+  yy = s * y * y;   yz = s * y * z;   wy = s * w * y;
+  zz = s * z * z;   xz = s * x * z;   wz = s * w * z;
+
+  dest[0][0] = 1.0f - yy - zz;
+  dest[1][1] = 1.0f - xx - zz;
+  dest[2][2] = 1.0f - xx - yy;
+
+  dest[0][1] = xy + wz;
+  dest[1][2] = yz + wx;
+  dest[2][0] = xz + wy;
+
+  dest[1][0] = xy - wz;
+  dest[2][1] = yz - wx;
+  dest[0][2] = xz - wy;
+}
+
+/*!
+ * @brief convert quaternion to mat3 (transposed)
+ *
+ * @param[in]   q     quaternion
+ * @param[out]  dest  result matrix
+ */
+CGLM_INLINE
+void
+glm_quat_mat3t(versor q, mat3 dest) {
+  float w, x, y, z,
+        xx, yy, zz,
+        xy, yz, xz,
+        wx, wy, wz, norm, s;
+
+  norm = glm_quat_norm(q);
+  s    = norm > 0.0f ? 2.0f / norm : 0.0f;
+
+  x = q[0];
+  y = q[1];
+  z = q[2];
+  w = q[3];
+
+  xx = s * x * x;   xy = s * x * y;   wx = s * w * x;
+  yy = s * y * y;   yz = s * y * z;   wy = s * w * y;
+  zz = s * z * z;   xz = s * x * z;   wz = s * w * z;
+
+  dest[0][0] = 1.0f - yy - zz;
+  dest[1][1] = 1.0f - xx - zz;
+  dest[2][2] = 1.0f - xx - yy;
+
+  dest[1][0] = xy + wz;
+  dest[2][1] = yz + wx;
+  dest[0][2] = xz + wy;
+
+  dest[0][1] = xy - wz;
+  dest[1][2] = yz - wx;
+  dest[2][0] = xz - wy;
+}
+
+/*!
+ * @brief interpolates between two quaternions
+ *        using linear interpolation (LERP)
+ *
+ * @param[in]   from  from
+ * @param[in]   to    to
+ * @param[in]   t     interpolant (amount)
+ * @param[out]  dest  result quaternion
+ */
+CGLM_INLINE
+void
+glm_quat_lerp(versor from, versor to, float t, versor dest) {
+  glm_vec4_lerp(from, to, t, dest);
+}
+
+/*!
+ * @brief interpolates between two quaternions
+ *        using linear interpolation (LERP)
+ *
+ * @param[in]   from  from
+ * @param[in]   to    to
+ * @param[in]   t     interpolant (amount) clamped between 0 and 1
+ * @param[out]  dest  result quaternion
+ */
+CGLM_INLINE
+void
+glm_quat_lerpc(versor from, versor to, float t, versor dest) {
+  glm_vec4_lerpc(from, to, t, dest);
+}
+
+/*!
+ * @brief interpolates between two quaternions
+ *        using spherical linear interpolation (SLERP)
+ *
+ * @param[in]   from  from
+ * @param[in]   to    to
+ * @param[in]   t     amout
+ * @param[out]  dest  result quaternion
+ */
+CGLM_INLINE
+void
+glm_quat_slerp(versor from, versor to, float t, versor dest) {
+  CGLM_ALIGN(16) vec4 q1, q2;
+  float cosTheta, sinTheta, angle;
+
+  cosTheta = glm_quat_dot(from, to);
+  glm_quat_copy(from, q1);
+
+  if (fabsf(cosTheta) >= 1.0f) {
+    glm_quat_copy(q1, dest);
     return;
   }
 
-  sinTheta = sqrt(1.0f - cosTheta * cosTheta);
+  if (cosTheta < 0.0f) {
+    glm_vec4_negate(q1);
+    cosTheta = -cosTheta;
+  }
 
-  c = 1.0f - t;
+  sinTheta = sqrtf(1.0f - cosTheta * cosTheta);
 
-  /* LERP */
-  /* TODO: FLT_EPSILON vs 0.001? */
-  if (sinTheta < 0.001f) {
-    dest[0] = c * q[0] + t * r[0];
-    dest[1] = c * q[1] + t * r[1];
-    dest[2] = c * q[2] + t * r[2];
-    dest[3] = c * q[3] + t * r[3];
+  /* LERP to avoid zero division */
+  if (fabsf(sinTheta) < 0.001f) {
+    glm_quat_lerp(from, to, t, dest);
     return;
   }
 
   /* SLERP */
   angle = acosf(cosTheta);
-  a = sinf(c * angle);
-  b = sinf(t * angle);
+  glm_vec4_scale(q1, sinf((1.0f - t) * angle), q1);
+  glm_vec4_scale(to, sinf(t * angle), q2);
 
-  dest[0] = (q[0] * a + r[0] * b) / sinTheta;
-  dest[1] = (q[1] * a + r[1] * b) / sinTheta;
-  dest[2] = (q[2] * a + r[2] * b) / sinTheta;
-  dest[3] = (q[3] * a + r[3] * b) / sinTheta;
-#endif
+  glm_vec4_add(q1, q2, q1);
+  glm_vec4_scale(q1, 1.0f / sinTheta, dest);
+}
+
+/*!
+ * @brief creates view matrix using quaternion as camera orientation
+ *
+ * @param[in]   eye   eye
+ * @param[in]   ori   orientation in world space as quaternion
+ * @param[out]  dest  view matrix
+ */
+CGLM_INLINE
+void
+glm_quat_look(vec3 eye, versor ori, mat4 dest) {
+  /* orientation */
+  glm_quat_mat4t(ori, dest);
+
+  /* translate */
+  glm_mat4_mulv3(dest, eye, 1.0f, dest[3]);
+  glm_vec3_negate(dest[3]);
+}
+
+/*!
+ * @brief creates look rotation quaternion
+ *
+ * @param[in]   dir   direction to look
+ * @param[in]   up    up vector
+ * @param[out]  dest  destination quaternion
+ */
+CGLM_INLINE
+void
+glm_quat_for(vec3 dir, vec3 up, versor dest) {
+  CGLM_ALIGN_MAT mat3 m;
+
+  glm_vec3_normalize_to(dir, m[2]); 
+
+  /* No need to negate in LH, but we use RH here */
+  glm_vec3_negate(m[2]);
+  
+  glm_vec3_crossn(up, m[2], m[0]);
+  glm_vec3_cross(m[2], m[0], m[1]);
+
+  glm_mat3_quat(m, dest);
+}
+
+/*!
+ * @brief creates look rotation quaternion using source and
+ *        destination positions p suffix stands for position
+ *
+ * @param[in]   from  source point
+ * @param[in]   to    destination point
+ * @param[in]   up    up vector
+ * @param[out]  dest  destination quaternion
+ */
+CGLM_INLINE
+void
+glm_quat_forp(vec3 from, vec3 to, vec3 up, versor dest) {
+  CGLM_ALIGN(8) vec3 dir;
+  glm_vec3_sub(to, from, dir);
+  glm_quat_for(dir, up, dest);
+}
+
+/*!
+ * @brief rotate vector using using quaternion
+ *
+ * @param[in]   q     quaternion
+ * @param[in]   v     vector to rotate
+ * @param[out]  dest  rotated vector
+ */
+CGLM_INLINE
+void
+glm_quat_rotatev(versor q, vec3 v, vec3 dest) {
+  CGLM_ALIGN(16) versor p;
+  CGLM_ALIGN(8)  vec3   u, v1, v2;
+  float s;
+
+  glm_quat_normalize_to(q, p);
+  glm_quat_imag(p, u);
+  s = glm_quat_real(p);
+
+  glm_vec3_scale(u, 2.0f * glm_vec3_dot(u, v), v1);
+  glm_vec3_scale(v, s * s - glm_vec3_dot(u, u), v2);
+  glm_vec3_add(v1, v2, v1);
+
+  glm_vec3_cross(u, v, v2);
+  glm_vec3_scale(v2, 2.0f * s, v2);
+
+  glm_vec3_add(v1, v2, dest);
+}
+
+/*!
+ * @brief rotate existing transform matrix using quaternion
+ *
+ * @param[in]   m     existing transform matrix
+ * @param[in]   q     quaternion
+ * @param[out]  dest  rotated matrix/transform
+ */
+CGLM_INLINE
+void
+glm_quat_rotate(mat4 m, versor q, mat4 dest) {
+  CGLM_ALIGN_MAT mat4 rot;
+  glm_quat_mat4(q, rot);
+  glm_mul_rot(m, rot, dest);
+}
+
+/*!
+ * @brief rotate existing transform matrix using quaternion at pivot point
+ *
+ * @param[in, out]   m     existing transform matrix
+ * @param[in]        q     quaternion
+ * @param[out]       pivot pivot
+ */
+CGLM_INLINE
+void
+glm_quat_rotate_at(mat4 m, versor q, vec3 pivot) {
+  CGLM_ALIGN(8) vec3 pivotInv;
+
+  glm_vec3_negate_to(pivot, pivotInv);
+
+  glm_translate(m, pivot);
+  glm_quat_rotate(m, q, m);
+  glm_translate(m, pivotInv);
+}
+
+/*!
+ * @brief rotate NEW transform matrix using quaternion at pivot point
+ *
+ * this creates rotation matrix, it assumes you don't have a matrix
+ *
+ * this should work faster than glm_quat_rotate_at because it reduces
+ * one glm_translate.
+ *
+ * @param[out]  m     existing transform matrix
+ * @param[in]   q     quaternion
+ * @param[in]   pivot pivot
+ */
+CGLM_INLINE
+void
+glm_quat_rotate_atm(mat4 m, versor q, vec3 pivot) {
+  CGLM_ALIGN(8) vec3 pivotInv;
+
+  glm_vec3_negate_to(pivot, pivotInv);
+
+  glm_translate_make(m, pivot);
+  glm_quat_rotate(m, q, m);
+  glm_translate(m, pivotInv);
 }
 
 #endif /* cglm_quat_h */

include/cglm/simd/arm.h

@@ -0,0 +1,83 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglm_simd_arm_h
+#define cglm_simd_arm_h
+#include "intrin.h"
+#ifdef CGLM_SIMD_ARM
+
+#define glmm_load(p)      vld1q_f32(p)
+#define glmm_store(p, a)  vst1q_f32(p, a)
+
+static inline
+float32x4_t
+glmm_abs(float32x4_t v) {
+  return vabsq_f32(v);
+}
+
+static inline
+float
+glmm_hadd(float32x4_t v) {
+#if defined(__aarch64__)
+  return vaddvq_f32(v);
+#else
+  v = vaddq_f32(v, vrev64q_f32(v));
+  v = vaddq_f32(v, vcombine_f32(vget_high_f32(v), vget_low_f32(v)));
+  return vgetq_lane_f32(v, 0);
+#endif
+}
+
+static inline
+float
+glmm_hmin(float32x4_t v) {
+  float32x2_t t;
+  t = vpmin_f32(vget_low_f32(v), vget_high_f32(v));
+  t = vpmin_f32(t, t);
+  return vget_lane_f32(t, 0);
+}
+
+static inline
+float
+glmm_hmax(float32x4_t v) {
+  float32x2_t t;
+  t = vpmax_f32(vget_low_f32(v), vget_high_f32(v));
+  t = vpmax_f32(t, t);
+  return vget_lane_f32(t, 0);
+}
+
+static inline
+float
+glmm_dot(float32x4_t a, float32x4_t b) {
+  return glmm_hadd(vmulq_f32(a, b));
+}
+
+static inline
+float
+glmm_norm(float32x4_t a) {
+  return sqrtf(glmm_dot(a, a));
+}
+
+static inline
+float
+glmm_norm2(float32x4_t a) {
+  return glmm_dot(a, a);
+}
+
+static inline
+float
+glmm_norm_one(float32x4_t a) {
+  return glmm_hadd(glmm_abs(a));
+}
+
+static inline
+float
+glmm_norm_inf(float32x4_t a) {
+  return glmm_hmax(glmm_abs(a));
+}
+
+#endif
+#endif /* cglm_simd_arm_h */

include/cglm/simd/avx/affine.h

@@ -21,27 +21,30 @@ glm_mul_avx(mat4 m1, mat4 m2, mat4 dest) {
 
   __m256 y0, y1, y2, y3, y4, y5, y6, y7, y8, y9;
 
-  y0 = _mm256_load_ps(m2[0]); /* h g f e d c b a */
-  y1 = _mm256_load_ps(m2[2]); /* p o n m l k j i */
+  y0 = glmm_load256(m2[0]); /* h g f e d c b a */
+  y1 = glmm_load256(m2[2]); /* p o n m l k j i */
 
-  y2 = _mm256_load_ps(m1[0]); /* h g f e d c b a */
-  y3 = _mm256_load_ps(m1[2]); /* p o n m l k j i */
+  y2 = glmm_load256(m1[0]); /* h g f e d c b a */
+  y3 = glmm_load256(m1[2]); /* p o n m l k j i */
 
-  y4 = _mm256_permute2f128_ps(y2, y2, 0b00000011); /* d c b a h g f e */
-  y5 = _mm256_permute2f128_ps(y3, y3, 0b00000000); /* l k j i l k j i */
+  /* 0x03: 0b00000011 */
+  y4 = _mm256_permute2f128_ps(y2, y2, 0x03); /* d c b a h g f e */
+  y5 = _mm256_permute2f128_ps(y3, y3, 0x03); /* l k j i p o n m */
 
   /* f f f f a a a a */
-  /* g g g g c c c c */
+  /* h h h h c c c c */
   /* e e e e b b b b */
-  y7 = _mm256_permute_ps(y0, 0b10101010);
+  /* g g g g d d d d */
   y6 = _mm256_permutevar_ps(y0, _mm256_set_epi32(1, 1, 1, 1, 0, 0, 0, 0));
+  y7 = _mm256_permutevar_ps(y0, _mm256_set_epi32(3, 3, 3, 3, 2, 2, 2, 2));
   y8 = _mm256_permutevar_ps(y0, _mm256_set_epi32(0, 0, 0, 0, 1, 1, 1, 1));
+  y9 = _mm256_permutevar_ps(y0, _mm256_set_epi32(2, 2, 2, 2, 3, 3, 3, 3));
 
-  _mm256_store_ps(dest[0],
-                  _mm256_add_ps(_mm256_add_ps(_mm256_mul_ps(y2, y6),
-                                              _mm256_mul_ps(y4, y8)),
-                                _mm256_mul_ps(y5, y7)));
-
+  glmm_store256(dest[0],
+                _mm256_add_ps(_mm256_add_ps(_mm256_mul_ps(y2, y6),
+                                            _mm256_mul_ps(y3, y7)),
+                              _mm256_add_ps(_mm256_mul_ps(y4, y8),
+                                            _mm256_mul_ps(y5, y9))));
 
   /* n n n n i i i i */
   /* p p p p k k k k */
@@ -52,11 +55,11 @@ glm_mul_avx(mat4 m1, mat4 m2, mat4 dest) {
   y8 = _mm256_permutevar_ps(y1, _mm256_set_epi32(0, 0, 0, 0, 1, 1, 1, 1));
   y9 = _mm256_permutevar_ps(y1, _mm256_set_epi32(2, 2, 2, 2, 3, 3, 3, 3));
 
-  _mm256_store_ps(dest[2],
-                  _mm256_add_ps(_mm256_add_ps(_mm256_mul_ps(y2, y6),
-                                              _mm256_mul_ps(y3, y7)),
-                                _mm256_add_ps(_mm256_mul_ps(y4, y8),
-                                              _mm256_mul_ps(y5, y9))));
+  glmm_store256(dest[2],
+                _mm256_add_ps(_mm256_add_ps(_mm256_mul_ps(y2, y6),
+                                            _mm256_mul_ps(y3, y7)),
+                              _mm256_add_ps(_mm256_mul_ps(y4, y8),
+                                            _mm256_mul_ps(y5, y9))));
 }
 
 #endif

include/cglm/simd/avx/mat4.h

@@ -21,14 +21,15 @@ glm_mat4_mul_avx(mat4 m1, mat4 m2, mat4 dest) {
 
   __m256 y0, y1, y2, y3, y4, y5, y6, y7, y8, y9;
 
-  y0 = _mm256_load_ps(m2[0]); /* h g f e d c b a */
-  y1 = _mm256_load_ps(m2[2]); /* p o n m l k j i */
+  y0 = glmm_load256(m2[0]); /* h g f e d c b a */
+  y1 = glmm_load256(m2[2]); /* p o n m l k j i */
 
-  y2 = _mm256_load_ps(m1[0]); /* h g f e d c b a */
-  y3 = _mm256_load_ps(m1[2]); /* p o n m l k j i */
+  y2 = glmm_load256(m1[0]); /* h g f e d c b a */
+  y3 = glmm_load256(m1[2]); /* p o n m l k j i */
 
-  y4 = _mm256_permute2f128_ps(y2, y2, 0b00000011); /* d c b a h g f e */
-  y5 = _mm256_permute2f128_ps(y3, y3, 0b00000011); /* l k j i p o n m */
+  /* 0x03: 0b00000011 */
+  y4 = _mm256_permute2f128_ps(y2, y2, 0x03); /* d c b a h g f e */
+  y5 = _mm256_permute2f128_ps(y3, y3, 0x03); /* l k j i p o n m */
 
   /* f f f f a a a a */
   /* h h h h c c c c */
@@ -39,11 +40,11 @@ glm_mat4_mul_avx(mat4 m1, mat4 m2, mat4 dest) {
   y8 = _mm256_permutevar_ps(y0, _mm256_set_epi32(0, 0, 0, 0, 1, 1, 1, 1));
   y9 = _mm256_permutevar_ps(y0, _mm256_set_epi32(2, 2, 2, 2, 3, 3, 3, 3));
 
-  _mm256_store_ps(dest[0],
-                  _mm256_add_ps(_mm256_add_ps(_mm256_mul_ps(y2, y6),
-                                              _mm256_mul_ps(y3, y7)),
-                                _mm256_add_ps(_mm256_mul_ps(y4, y8),
-                                              _mm256_mul_ps(y5, y9))));
+  glmm_store256(dest[0],
+                _mm256_add_ps(_mm256_add_ps(_mm256_mul_ps(y2, y6),
+                                            _mm256_mul_ps(y3, y7)),
+                              _mm256_add_ps(_mm256_mul_ps(y4, y8),
+                                            _mm256_mul_ps(y5, y9))));
 
   /* n n n n i i i i */
   /* p p p p k k k k */
@@ -54,11 +55,11 @@ glm_mat4_mul_avx(mat4 m1, mat4 m2, mat4 dest) {
   y8 = _mm256_permutevar_ps(y1, _mm256_set_epi32(0, 0, 0, 0, 1, 1, 1, 1));
   y9 = _mm256_permutevar_ps(y1, _mm256_set_epi32(2, 2, 2, 2, 3, 3, 3, 3));
 
-  _mm256_store_ps(dest[2],
-                  _mm256_add_ps(_mm256_add_ps(_mm256_mul_ps(y2, y6),
-                                              _mm256_mul_ps(y3, y7)),
-                                _mm256_add_ps(_mm256_mul_ps(y4, y8),
-                                              _mm256_mul_ps(y5, y9))));
+  glmm_store256(dest[2],
+                _mm256_add_ps(_mm256_add_ps(_mm256_mul_ps(y2, y6),
+                                            _mm256_mul_ps(y3, y7)),
+                              _mm256_add_ps(_mm256_mul_ps(y4, y8),
+                                            _mm256_mul_ps(y5, y9))));
 }
 
 #endif

include/cglm/simd/intrin.h

@@ -8,45 +8,83 @@
 #ifndef cglm_intrin_h
 #define cglm_intrin_h
 
-#if defined( _WIN32 )
+#if defined( _MSC_VER )
 #  if (defined(_M_AMD64) || defined(_M_X64)) || _M_IX86_FP == 2
-#    define __SSE2__
+#    ifndef __SSE2__
+#      define __SSE2__
+#    endif
 #  elif _M_IX86_FP == 1
-#    define __SSE__
+#    ifndef __SSE__
+#      define __SSE__
+#    endif
+#  endif
+/* do not use alignment for older visual studio versions */
+#  if _MSC_VER < 1913     /* Visual Studio 2017 version 15.6 */
+#    define CGLM_ALL_UNALIGNED
 #  endif
 #endif
 
 #if defined( __SSE__ ) || defined( __SSE2__ )
 #  include <xmmintrin.h>
 #  include <emmintrin.h>
-
-/* float */
-#  define _mm_shuffle1_ps(a, z, y, x, w)                                       \
-     _mm_shuffle_ps(a, a, _MM_SHUFFLE(z, y, x, w))
-
-#  define _mm_shuffle1_ps1(a, x)                                               \
-     _mm_shuffle_ps(a, a, _MM_SHUFFLE(x, x, x, x))
-
-#  define _mm_shuffle2_ps(a, b, z0, y0, x0, w0, z1, y1, x1, w1)                \
-     _mm_shuffle1_ps(_mm_shuffle_ps(a, b, _MM_SHUFFLE(z0, y0, x0, w0)),        \
-                                    z1, y1, x1, w1)
+#  define CGLM_SSE_FP 1
+#  ifndef CGLM_SIMD_x86
+#    define CGLM_SIMD_x86
+#  endif
 #endif
 
-/* x86, x64 */
-#if defined( __SSE__ ) || defined( __SSE2__ )
-#  define CGLM_SSE_FP 1
+#if defined(__SSE3__)
+#  include <x86intrin.h>
+#  ifndef CGLM_SIMD_x86
+#    define CGLM_SIMD_x86
+#  endif
+#endif
+
+#if defined(__SSE4_1__)
+#  include <smmintrin.h>
+#  ifndef CGLM_SIMD_x86
+#    define CGLM_SIMD_x86
+#  endif
+#endif
+
+#if defined(__SSE4_2__)
+#  include <nmmintrin.h>
+#  ifndef CGLM_SIMD_x86
+#    define CGLM_SIMD_x86
+#  endif
 #endif
 
 #ifdef __AVX__
+#  include <immintrin.h>
 #  define CGLM_AVX_FP 1
+#  ifndef CGLM_SIMD_x86
+#    define CGLM_SIMD_x86
+#  endif
 #endif
 
 /* ARM Neon */
-#if defined(__ARM_NEON) && defined(__ARM_NEON_FP)
+#if defined(__ARM_NEON)
 #  include <arm_neon.h>
-#  define CGLM_NEON_FP 1
-#else
-#  undef  CGLM_NEON_FP
+#  if defined(__ARM_NEON_FP)
+#    define CGLM_NEON_FP 1
+#    ifndef CGLM_SIMD_ARM
+#      define CGLM_SIMD_ARM
+#    endif
+#  endif
+#endif
+
+#if defined(CGLM_SIMD_x86) || defined(CGLM_NEON_FP)
+#  ifndef CGLM_SIMD
+#    define CGLM_SIMD
+#  endif
+#endif
+
+#if defined(CGLM_SIMD_x86)
+#  include "x86.h"
+#endif
+
+#if defined(CGLM_SIMD_ARM)
+#  include "arm.h"
 #endif
 
 #endif /* cglm_intrin_h */

include/cglm/simd/sse2/affine.h

@@ -18,35 +18,67 @@ glm_mul_sse2(mat4 m1, mat4 m2, mat4 dest) {
   /* D = R * L (Column-Major) */
   __m128 l0, l1, l2, l3, r;
 
-  l0 = _mm_load_ps(m1[0]);
-  l1 = _mm_load_ps(m1[1]);
-  l2 = _mm_load_ps(m1[2]);
-  l3 = _mm_load_ps(m1[3]);
+  l0 = glmm_load(m1[0]);
+  l1 = glmm_load(m1[1]);
+  l2 = glmm_load(m1[2]);
+  l3 = glmm_load(m1[3]);
 
-  r = _mm_load_ps(m2[0]);
-  _mm_store_ps(dest[0],
-               _mm_add_ps(_mm_add_ps(_mm_mul_ps(_mm_shuffle1_ps1(r, 0), l0),
-                                     _mm_mul_ps(_mm_shuffle1_ps1(r, 1), l1)),
-                          _mm_mul_ps(_mm_shuffle1_ps1(r, 2), l2)));
+  r = glmm_load(m2[0]);
+  glmm_store(dest[0],
+             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 0), l0),
+                                   _mm_mul_ps(glmm_shuff1x(r, 1), l1)),
+                        _mm_mul_ps(glmm_shuff1x(r, 2), l2)));
 
-  r = _mm_load_ps(m2[1]);
-  _mm_store_ps(dest[1],
-               _mm_add_ps(_mm_add_ps(_mm_mul_ps(_mm_shuffle1_ps1(r, 0), l0),
-                                     _mm_mul_ps(_mm_shuffle1_ps1(r, 1), l1)),
-                          _mm_mul_ps(_mm_shuffle1_ps1(r, 2), l2)));
+  r = glmm_load(m2[1]);
+  glmm_store(dest[1],
+             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 0), l0),
+                                   _mm_mul_ps(glmm_shuff1x(r, 1), l1)),
+                        _mm_mul_ps(glmm_shuff1x(r, 2), l2)));
 
-  r = _mm_load_ps(m2[2]);
-  _mm_store_ps(dest[2],
-               _mm_add_ps(_mm_add_ps(_mm_mul_ps(_mm_shuffle1_ps1(r, 0), l0),
-                                     _mm_mul_ps(_mm_shuffle1_ps1(r, 1), l1)),
-                          _mm_mul_ps(_mm_shuffle1_ps1(r, 2), l2)));
+  r = glmm_load(m2[2]);
+  glmm_store(dest[2],
+             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 0), l0),
+                                   _mm_mul_ps(glmm_shuff1x(r, 1), l1)),
+                        _mm_mul_ps(glmm_shuff1x(r, 2), l2)));
 
-  r = _mm_load_ps(m2[3]);
-  _mm_store_ps(dest[3],
-               _mm_add_ps(_mm_add_ps(_mm_mul_ps(_mm_shuffle1_ps1(r, 0), l0),
-                                     _mm_mul_ps(_mm_shuffle1_ps1(r, 1), l1)),
-                          _mm_add_ps(_mm_mul_ps(_mm_shuffle1_ps1(r, 2), l2),
-                                     _mm_mul_ps(_mm_shuffle1_ps1(r, 3), l3))));
+  r = glmm_load(m2[3]);
+  glmm_store(dest[3],
+             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 0), l0),
+                                   _mm_mul_ps(glmm_shuff1x(r, 1), l1)),
+                        _mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 2), l2),
+                                   _mm_mul_ps(glmm_shuff1x(r, 3), l3))));
+}
+
+CGLM_INLINE
+void
+glm_mul_rot_sse2(mat4 m1, mat4 m2, mat4 dest) {
+  /* D = R * L (Column-Major) */
+  __m128 l0, l1, l2, l3, r;
+
+  l0 = glmm_load(m1[0]);
+  l1 = glmm_load(m1[1]);
+  l2 = glmm_load(m1[2]);
+  l3 = glmm_load(m1[3]);
+
+  r = glmm_load(m2[0]);
+  glmm_store(dest[0],
+             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 0), l0),
+                                   _mm_mul_ps(glmm_shuff1x(r, 1), l1)),
+                        _mm_mul_ps(glmm_shuff1x(r, 2), l2)));
+
+  r = glmm_load(m2[1]);
+  glmm_store(dest[1],
+             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 0), l0),
+                                   _mm_mul_ps(glmm_shuff1x(r, 1), l1)),
+                        _mm_mul_ps(glmm_shuff1x(r, 2), l2)));
+
+  r = glmm_load(m2[2]);
+  glmm_store(dest[2],
+             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 0), l0),
+                                   _mm_mul_ps(glmm_shuff1x(r, 1), l1)),
+                        _mm_mul_ps(glmm_shuff1x(r, 2), l2)));
+
+  glmm_store(dest[3], l3);
 }
 
 CGLM_INLINE
@@ -54,25 +86,25 @@ void
 glm_inv_tr_sse2(mat4 mat) {
   __m128 r0, r1, r2, r3, x0, x1;
 
-  r0 = _mm_load_ps(mat[0]);
-  r1 = _mm_load_ps(mat[1]);
-  r2 = _mm_load_ps(mat[2]);
-  r3 = _mm_load_ps(mat[3]);
-  x1  = _mm_set_ps(1.0f, 0.0f, 0.0f, 0.0f);
+  r0 = glmm_load(mat[0]);
+  r1 = glmm_load(mat[1]);
+  r2 = glmm_load(mat[2]);
+  r3 = glmm_load(mat[3]);
+  x1 = _mm_set_ps(1.0f, 0.0f, 0.0f, 0.0f);
 
   _MM_TRANSPOSE4_PS(r0, r1, r2, x1);
 
-  x0 = _mm_add_ps(_mm_mul_ps(r0, _mm_shuffle1_ps(r3, 0, 0, 0, 0)),
-                  _mm_mul_ps(r1, _mm_shuffle1_ps(r3, 1, 1, 1, 1)));
-  x0 = _mm_add_ps(x0, _mm_mul_ps(r2, _mm_shuffle1_ps(r3, 2, 2, 2, 2)));
+  x0 = _mm_add_ps(_mm_mul_ps(r0, glmm_shuff1(r3, 0, 0, 0, 0)),
+                  _mm_mul_ps(r1, glmm_shuff1(r3, 1, 1, 1, 1)));
+  x0 = _mm_add_ps(x0, _mm_mul_ps(r2, glmm_shuff1(r3, 2, 2, 2, 2)));
   x0 = _mm_xor_ps(x0, _mm_set1_ps(-0.f));
 
   x0 = _mm_add_ps(x0, x1);
 
-  _mm_store_ps(mat[0], r0);
-  _mm_store_ps(mat[1], r1);
-  _mm_store_ps(mat[2], r2);
-  _mm_store_ps(mat[3], x0);
+  glmm_store(mat[0], r0);
+  glmm_store(mat[1], r1);
+  glmm_store(mat[2], r2);
+  glmm_store(mat[3], x0);
 }
 
 #endif

include/cglm/simd/sse2/mat2.h

@@ -0,0 +1,45 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglm_mat2_sse_h
+#define cglm_mat2_sse_h
+#if defined( __SSE__ ) || defined( __SSE2__ )
+
+#include "../../common.h"
+#include "../intrin.h"
+
+CGLM_INLINE
+void
+glm_mat2_mul_sse2(mat2 m1, mat2 m2, mat2 dest) {
+  __m128 x0, x1, x2;
+
+  x1 = glmm_load(m1[0]); /* d c b a */
+  x2 = glmm_load(m2[0]); /* h g f e */
+
+  /*
+   dest[0][0] = a * e + c * f;
+   dest[0][1] = b * e + d * f;
+   dest[1][0] = a * g + c * h;
+   dest[1][1] = b * g + d * h;
+   */
+  x0 = _mm_mul_ps(_mm_movelh_ps(x1, x1), glmm_shuff1(x2, 2, 2, 0, 0));
+  x1 = _mm_mul_ps(_mm_movehl_ps(x1, x1), glmm_shuff1(x2, 3, 3, 1, 1));
+  x1 = _mm_add_ps(x0, x1);
+
+  glmm_store(dest[0], x1);
+}
+
+CGLM_INLINE
+void
+glm_mat2_transp_sse2(mat2 m, mat2 dest) {
+  /* d c b a */
+  /* d b c a */
+  glmm_store(dest[0], glmm_shuff1(glmm_load(m[0]), 3, 1, 2, 0));
+}
+
+#endif
+#endif /* cglm_mat2_sse_h */

include/cglm/simd/sse2/mat3.h

@@ -27,27 +27,25 @@ glm_mat3_mul_sse2(mat3 m1, mat3 m2, mat3 dest) {
   r1 = _mm_loadu_ps(&m2[1][1]);
   r2 = _mm_set1_ps(m2[2][2]);
 
-  x1 = _mm_shuffle2_ps(l0, l1, 1, 0, 3, 3, 0, 3, 2, 0);
-  x2 = _mm_shuffle2_ps(l1, l2, 0, 0, 3, 2, 0, 2, 1, 0);
+  x1 = glmm_shuff2(l0, l1, 1, 0, 3, 3, 0, 3, 2, 0);
+  x2 = glmm_shuff2(l1, l2, 0, 0, 3, 2, 0, 2, 1, 0);
 
-  x0 = _mm_add_ps(_mm_mul_ps(_mm_shuffle1_ps(l0, 0, 2, 1, 0),
-                             _mm_shuffle1_ps(r0, 3, 0, 0, 0)),
-                  _mm_mul_ps(x1,
-                             _mm_shuffle2_ps(r0, r1, 0, 0, 1, 1, 2, 0, 0, 0)));
+  x0 = _mm_add_ps(_mm_mul_ps(glmm_shuff1(l0, 0, 2, 1, 0),
+                             glmm_shuff1(r0, 3, 0, 0, 0)),
+                  _mm_mul_ps(x1, glmm_shuff2(r0, r1, 0, 0, 1, 1, 2, 0, 0, 0)));
 
   x0 = _mm_add_ps(x0,
-                  _mm_mul_ps(x2,
-                             _mm_shuffle2_ps(r0, r1, 1, 1, 2, 2, 2, 0, 0, 0)));
+                  _mm_mul_ps(x2, glmm_shuff2(r0, r1, 1, 1, 2, 2, 2, 0, 0, 0)));
 
   _mm_storeu_ps(dest[0], x0);
 
-  x0 = _mm_add_ps(_mm_mul_ps(_mm_shuffle1_ps(l0, 1, 0, 2, 1),
+  x0 = _mm_add_ps(_mm_mul_ps(glmm_shuff1(l0, 1, 0, 2, 1),
                              _mm_shuffle_ps(r0, r1, _MM_SHUFFLE(2, 2, 3, 3))),
-                  _mm_mul_ps(_mm_shuffle1_ps(x1, 1, 0, 2, 1),
-                             _mm_shuffle1_ps(r1, 3, 3, 0, 0)));
+                  _mm_mul_ps(glmm_shuff1(x1, 1, 0, 2, 1),
+                             glmm_shuff1(r1, 3, 3, 0, 0)));
 
   x0 = _mm_add_ps(x0,
-                  _mm_mul_ps(_mm_shuffle1_ps(x2, 1, 0, 2, 1),
+                  _mm_mul_ps(glmm_shuff1(x2, 1, 0, 2, 1),
                              _mm_shuffle_ps(r1, r2, _MM_SHUFFLE(0, 0, 1, 1))));
 
   _mm_storeu_ps(&dest[1][1], x0);

include/cglm/simd/sse2/mat4.h

@@ -12,34 +12,36 @@
 #include "../../common.h"
 #include "../intrin.h"
 
+#define glm_mat4_inv_precise_sse2(mat, dest) glm_mat4_inv_sse2(mat, dest)
+
 CGLM_INLINE
 void
-glm_mat4_scale_sse2(mat4 m, float s){
+glm_mat4_scale_sse2(mat4 m, float s) {
   __m128 x0;
   x0 = _mm_set1_ps(s);
 
-  _mm_store_ps(m[0], _mm_mul_ps(_mm_load_ps(m[0]), x0));
-  _mm_store_ps(m[1], _mm_mul_ps(_mm_load_ps(m[1]), x0));
-  _mm_store_ps(m[2], _mm_mul_ps(_mm_load_ps(m[2]), x0));
-  _mm_store_ps(m[3], _mm_mul_ps(_mm_load_ps(m[3]), x0));
+  glmm_store(m[0], _mm_mul_ps(glmm_load(m[0]), x0));
+  glmm_store(m[1], _mm_mul_ps(glmm_load(m[1]), x0));
+  glmm_store(m[2], _mm_mul_ps(glmm_load(m[2]), x0));
+  glmm_store(m[3], _mm_mul_ps(glmm_load(m[3]), x0));
 }
 
 CGLM_INLINE
 void
-glm_mat4_transp_sse2(mat4 m, mat4 dest){
+glm_mat4_transp_sse2(mat4 m, mat4 dest) {
   __m128 r0, r1, r2, r3;
 
-  r0 = _mm_load_ps(m[0]);
-  r1 = _mm_load_ps(m[1]);
-  r2 = _mm_load_ps(m[2]);
-  r3 = _mm_load_ps(m[3]);
+  r0 = glmm_load(m[0]);
+  r1 = glmm_load(m[1]);
+  r2 = glmm_load(m[2]);
+  r3 = glmm_load(m[3]);
 
   _MM_TRANSPOSE4_PS(r0, r1, r2, r3);
 
-  _mm_store_ps(dest[0], r0);
-  _mm_store_ps(dest[1], r1);
-  _mm_store_ps(dest[2], r2);
-  _mm_store_ps(dest[3], r3);
+  glmm_store(dest[0], r0);
+  glmm_store(dest[1], r1);
+  glmm_store(dest[2], r2);
+  glmm_store(dest[3], r3);
 }
 
 CGLM_INLINE
@@ -49,36 +51,36 @@ glm_mat4_mul_sse2(mat4 m1, mat4 m2, mat4 dest) {
 
   __m128 l0, l1, l2, l3, r;
 
-  l0 = _mm_load_ps(m1[0]);
-  l1 = _mm_load_ps(m1[1]);
-  l2 = _mm_load_ps(m1[2]);
-  l3 = _mm_load_ps(m1[3]);
+  l0 = glmm_load(m1[0]);
+  l1 = glmm_load(m1[1]);
+  l2 = glmm_load(m1[2]);
+  l3 = glmm_load(m1[3]);
 
-  r = _mm_load_ps(m2[0]);
-  _mm_store_ps(dest[0],
-               _mm_add_ps(_mm_add_ps(_mm_mul_ps(_mm_shuffle1_ps1(r, 0), l0),
-                                     _mm_mul_ps(_mm_shuffle1_ps1(r, 1), l1)),
-                          _mm_add_ps(_mm_mul_ps(_mm_shuffle1_ps1(r, 2), l2),
-                                     _mm_mul_ps(_mm_shuffle1_ps1(r, 3), l3))));
-  r = _mm_load_ps(m2[1]);
-  _mm_store_ps(dest[1],
-               _mm_add_ps(_mm_add_ps(_mm_mul_ps(_mm_shuffle1_ps1(r, 0), l0),
-                                     _mm_mul_ps(_mm_shuffle1_ps1(r, 1), l1)),
-                          _mm_add_ps(_mm_mul_ps(_mm_shuffle1_ps1(r, 2), l2),
-                                     _mm_mul_ps(_mm_shuffle1_ps1(r, 3), l3))));
-  r = _mm_load_ps(m2[2]);
-  _mm_store_ps(dest[2],
-               _mm_add_ps(_mm_add_ps(_mm_mul_ps(_mm_shuffle1_ps1(r, 0), l0),
-                                     _mm_mul_ps(_mm_shuffle1_ps1(r, 1), l1)),
-                          _mm_add_ps(_mm_mul_ps(_mm_shuffle1_ps1(r, 2), l2),
-                                     _mm_mul_ps(_mm_shuffle1_ps1(r, 3), l3))));
+  r = glmm_load(m2[0]);
+  glmm_store(dest[0],
+             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 0), l0),
+                                   _mm_mul_ps(glmm_shuff1x(r, 1), l1)),
+                        _mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 2), l2),
+                                   _mm_mul_ps(glmm_shuff1x(r, 3), l3))));
+  r = glmm_load(m2[1]);
+  glmm_store(dest[1],
+             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 0), l0),
+                                   _mm_mul_ps(glmm_shuff1x(r, 1), l1)),
+                        _mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 2), l2),
+                                   _mm_mul_ps(glmm_shuff1x(r, 3), l3))));
+  r = glmm_load(m2[2]);
+  glmm_store(dest[2],
+             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 0), l0),
+                                   _mm_mul_ps(glmm_shuff1x(r, 1), l1)),
+                        _mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 2), l2),
+                                   _mm_mul_ps(glmm_shuff1x(r, 3), l3))));
 
-  r = _mm_load_ps(m2[3]);
-  _mm_store_ps(dest[3],
-               _mm_add_ps(_mm_add_ps(_mm_mul_ps(_mm_shuffle1_ps1(r, 0), l0),
-                                     _mm_mul_ps(_mm_shuffle1_ps1(r, 1), l1)),
-                          _mm_add_ps(_mm_mul_ps(_mm_shuffle1_ps1(r, 2), l2),
-                                     _mm_mul_ps(_mm_shuffle1_ps1(r, 3), l3))));
+  r = glmm_load(m2[3]);
+  glmm_store(dest[3],
+             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 0), l0),
+                                   _mm_mul_ps(glmm_shuff1x(r, 1), l1)),
+                        _mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 2), l2),
+                                   _mm_mul_ps(glmm_shuff1x(r, 3), l3))));
 }
 
 CGLM_INLINE
@@ -86,18 +88,14 @@ void
 glm_mat4_mulv_sse2(mat4 m, vec4 v, vec4 dest) {
   __m128 x0, x1, x2;
 
-  x0 = _mm_load_ps(v);
-  x1 = _mm_add_ps(_mm_mul_ps(_mm_load_ps(m[0]),
-                             _mm_shuffle1_ps1(x0, 0)),
-                  _mm_mul_ps(_mm_load_ps(m[1]),
-                             _mm_shuffle1_ps1(x0, 1)));
+  x0 = glmm_load(v);
+  x1 = _mm_add_ps(_mm_mul_ps(glmm_load(m[0]), glmm_shuff1x(x0, 0)),
+                  _mm_mul_ps(glmm_load(m[1]), glmm_shuff1x(x0, 1)));
 
-  x2 = _mm_add_ps(_mm_mul_ps(_mm_load_ps(m[2]),
-                             _mm_shuffle1_ps1(x0, 2)),
-                  _mm_mul_ps(_mm_load_ps(m[3]),
-                             _mm_shuffle1_ps1(x0, 3)));
+  x2 = _mm_add_ps(_mm_mul_ps(glmm_load(m[2]), glmm_shuff1x(x0, 2)),
+                  _mm_mul_ps(glmm_load(m[3]), glmm_shuff1x(x0, 3)));
 
-  _mm_store_ps(dest, _mm_add_ps(x1, x2));
+  glmm_store(dest, _mm_add_ps(x1, x2));
 }
 
 CGLM_INLINE
@@ -106,10 +104,10 @@ glm_mat4_det_sse2(mat4 mat) {
   __m128 r0, r1, r2, r3, x0, x1, x2;
 
   /* 127 <- 0, [square] det(A) = det(At) */
-  r0 = _mm_load_ps(mat[0]); /* d c b a */
-  r1 = _mm_load_ps(mat[1]); /* h g f e */
-  r2 = _mm_load_ps(mat[2]); /* l k j i */
-  r3 = _mm_load_ps(mat[3]); /* p o n m */
+  r0 = glmm_load(mat[0]); /* d c b a */
+  r1 = glmm_load(mat[1]); /* h g f e */
+  r2 = glmm_load(mat[2]); /* l k j i */
+  r3 = glmm_load(mat[3]); /* p o n m */
 
   /*
    t[1] = j * p - n * l;
@@ -117,20 +115,20 @@ glm_mat4_det_sse2(mat4 mat) {
    t[3] = i * p - m * l;
    t[4] = i * o - m * k;
    */
-  x0 = _mm_sub_ps(_mm_mul_ps(_mm_shuffle1_ps(r2, 0, 0, 1, 1),
-                             _mm_shuffle1_ps(r3, 2, 3, 2, 3)),
-                  _mm_mul_ps(_mm_shuffle1_ps(r3, 0, 0, 1, 1),
-                             _mm_shuffle1_ps(r2, 2, 3, 2, 3)));
+  x0 = _mm_sub_ps(_mm_mul_ps(glmm_shuff1(r2, 0, 0, 1, 1),
+                             glmm_shuff1(r3, 2, 3, 2, 3)),
+                  _mm_mul_ps(glmm_shuff1(r3, 0, 0, 1, 1),
+                             glmm_shuff1(r2, 2, 3, 2, 3)));
   /*
    t[0] = k * p - o * l;
    t[0] = k * p - o * l;
    t[5] = i * n - m * j;
    t[5] = i * n - m * j;
    */
-  x1 = _mm_sub_ps(_mm_mul_ps(_mm_shuffle1_ps(r2, 0, 0, 2, 2),
-                             _mm_shuffle1_ps(r3, 1, 1, 3, 3)),
-                  _mm_mul_ps(_mm_shuffle1_ps(r3, 0, 0, 2, 2),
-                             _mm_shuffle1_ps(r2, 1, 1, 3, 3)));
+  x1 = _mm_sub_ps(_mm_mul_ps(glmm_shuff1(r2, 0, 0, 2, 2),
+                             glmm_shuff1(r3, 1, 1, 3, 3)),
+                  _mm_mul_ps(glmm_shuff1(r3, 0, 0, 2, 2),
+                             glmm_shuff1(r2, 1, 1, 3, 3)));
 
   /*
      a * (f * t[0] - g * t[1] + h * t[2])
@@ -138,23 +136,147 @@ glm_mat4_det_sse2(mat4 mat) {
    + c * (e * t[1] - f * t[3] + h * t[5])
    - d * (e * t[2] - f * t[4] + g * t[5])
    */
-  x2 = _mm_sub_ps(_mm_mul_ps(_mm_shuffle1_ps(r1, 0, 0, 0, 1),
+  x2 = _mm_sub_ps(_mm_mul_ps(glmm_shuff1(r1, 0, 0, 0, 1),
                              _mm_shuffle_ps(x1, x0, _MM_SHUFFLE(1, 0, 0, 0))),
-                  _mm_mul_ps(_mm_shuffle1_ps(r1, 1, 1, 2, 2),
-                             _mm_shuffle1_ps(x0, 3, 2, 2, 0)));
+                  _mm_mul_ps(glmm_shuff1(r1, 1, 1, 2, 2),
+                             glmm_shuff1(x0, 3, 2, 2, 0)));
 
   x2 = _mm_add_ps(x2,
-                  _mm_mul_ps(_mm_shuffle1_ps(r1, 2, 3, 3, 3),
+                  _mm_mul_ps(glmm_shuff1(r1, 2, 3, 3, 3),
                              _mm_shuffle_ps(x0, x1, _MM_SHUFFLE(2, 2, 3, 1))));
   x2 = _mm_xor_ps(x2, _mm_set_ps(-0.f, 0.f, -0.f, 0.f));
 
   x0 = _mm_mul_ps(r0, x2);
-  x0 = _mm_add_ps(x0, _mm_shuffle1_ps(x0, 0, 1, 2, 3));
-  x0 = _mm_add_ps(x0, _mm_shuffle1_ps(x0, 1, 3, 3, 1));
+  x0 = _mm_add_ps(x0, glmm_shuff1(x0, 0, 1, 2, 3));
+  x0 = _mm_add_ps(x0, glmm_shuff1(x0, 1, 3, 3, 1));
 
   return _mm_cvtss_f32(x0);
 }
 
+CGLM_INLINE
+void
+glm_mat4_inv_fast_sse2(mat4 mat, mat4 dest) {
+  __m128 r0, r1, r2, r3,
+         v0, v1, v2, v3,
+         t0, t1, t2, t3, t4, t5,
+         x0, x1, x2, x3, x4, x5, x6, x7;
+
+  /* 127 <- 0 */
+  r0 = glmm_load(mat[0]); /* d c b a */
+  r1 = glmm_load(mat[1]); /* h g f e */
+  r2 = glmm_load(mat[2]); /* l k j i */
+  r3 = glmm_load(mat[3]); /* p o n m */
+
+  x0 = _mm_shuffle_ps(r2, r3, _MM_SHUFFLE(3, 2, 3, 2));  /* p o l k */
+  x1 = glmm_shuff1(x0, 1, 3, 3, 3);                      /* l p p p */
+  x2 = glmm_shuff1(x0, 0, 2, 2, 2);                      /* k o o o */
+  x0 = _mm_shuffle_ps(r2, r1, _MM_SHUFFLE(3, 3, 3, 3));  /* h h l l */
+  x3 = _mm_shuffle_ps(r2, r1, _MM_SHUFFLE(2, 2, 2, 2));  /* g g k k */
+
+  /* t1[0] = k * p - o * l;
+     t1[0] = k * p - o * l;
+     t2[0] = g * p - o * h;
+     t3[0] = g * l - k * h; */
+  t0 = _mm_sub_ps(_mm_mul_ps(x3, x1), _mm_mul_ps(x2, x0));
+
+  x4 = _mm_shuffle_ps(r2, r3, _MM_SHUFFLE(2, 1, 2, 1)); /* o n k j */
+  x4 = glmm_shuff1(x4, 0, 2, 2, 2);                     /* j n n n */
+  x5 = _mm_shuffle_ps(r2, r1, _MM_SHUFFLE(1, 1, 1, 1)); /* f f j j */
+
+  /* t1[1] = j * p - n * l;
+     t1[1] = j * p - n * l;
+     t2[1] = f * p - n * h;
+     t3[1] = f * l - j * h; */
+  t1 = _mm_sub_ps(_mm_mul_ps(x5, x1), _mm_mul_ps(x4, x0));
+
+  /* t1[2] = j * o - n * k
+     t1[2] = j * o - n * k;
+     t2[2] = f * o - n * g;
+     t3[2] = f * k - j * g; */
+  t2 = _mm_sub_ps(_mm_mul_ps(x5, x2), _mm_mul_ps(x4, x3));
+
+  x6 = _mm_shuffle_ps(r2, r1, _MM_SHUFFLE(0, 0, 0, 0)); /* e e i i */
+  x7 = glmm_shuff2(r3, r2, 0, 0, 0, 0, 2, 0, 0, 0);     /* i m m m */
+
+  /* t1[3] = i * p - m * l;
+     t1[3] = i * p - m * l;
+     t2[3] = e * p - m * h;
+     t3[3] = e * l - i * h; */
+  t3 = _mm_sub_ps(_mm_mul_ps(x6, x1), _mm_mul_ps(x7, x0));
+
+  /* t1[4] = i * o - m * k;
+     t1[4] = i * o - m * k;
+     t2[4] = e * o - m * g;
+     t3[4] = e * k - i * g; */
+  t4 = _mm_sub_ps(_mm_mul_ps(x6, x2), _mm_mul_ps(x7, x3));
+
+  /* t1[5] = i * n - m * j;
+     t1[5] = i * n - m * j;
+     t2[5] = e * n - m * f;
+     t3[5] = e * j - i * f; */
+  t5 = _mm_sub_ps(_mm_mul_ps(x6, x4), _mm_mul_ps(x7, x5));
+
+  x0 = glmm_shuff2(r1, r0, 0, 0, 0, 0, 2, 2, 2, 0); /* a a a e */
+  x1 = glmm_shuff2(r1, r0, 1, 1, 1, 1, 2, 2, 2, 0); /* b b b f */
+  x2 = glmm_shuff2(r1, r0, 2, 2, 2, 2, 2, 2, 2, 0); /* c c c g */
+  x3 = glmm_shuff2(r1, r0, 3, 3, 3, 3, 2, 2, 2, 0); /* d d d h */
+
+  /*
+   dest[0][0] =  f * t1[0] - g * t1[1] + h * t1[2];
+   dest[0][1] =-(b * t1[0] - c * t1[1] + d * t1[2]);
+   dest[0][2] =  b * t2[0] - c * t2[1] + d * t2[2];
+   dest[0][3] =-(b * t3[0] - c * t3[1] + d * t3[2]); */
+  v0 = _mm_add_ps(_mm_mul_ps(x3, t2),
+                  _mm_sub_ps(_mm_mul_ps(x1, t0),
+                             _mm_mul_ps(x2, t1)));
+  v0 = _mm_xor_ps(v0, _mm_set_ps(-0.f, 0.f, -0.f, 0.f));
+
+  /*
+   dest[1][0] =-(e * t1[0] - g * t1[3] + h * t1[4]);
+   dest[1][1] =  a * t1[0] - c * t1[3] + d * t1[4];
+   dest[1][2] =-(a * t2[0] - c * t2[3] + d * t2[4]);
+   dest[1][3] =  a * t3[0] - c * t3[3] + d * t3[4]; */
+  v1 = _mm_add_ps(_mm_mul_ps(x3, t4),
+                  _mm_sub_ps(_mm_mul_ps(x0, t0),
+                             _mm_mul_ps(x2, t3)));
+  v1 = _mm_xor_ps(v1, _mm_set_ps(0.f, -0.f, 0.f, -0.f));
+
+  /*
+   dest[2][0] =  e * t1[1] - f * t1[3] + h * t1[5];
+   dest[2][1] =-(a * t1[1] - b * t1[3] + d * t1[5]);
+   dest[2][2] =  a * t2[1] - b * t2[3] + d * t2[5];
+   dest[2][3] =-(a * t3[1] - b * t3[3] + d * t3[5]);*/
+  v2 = _mm_add_ps(_mm_mul_ps(x3, t5),
+                  _mm_sub_ps(_mm_mul_ps(x0, t1),
+                             _mm_mul_ps(x1, t3)));
+  v2 = _mm_xor_ps(v2, _mm_set_ps(-0.f, 0.f, -0.f, 0.f));
+
+  /*
+   dest[3][0] =-(e * t1[2] - f * t1[4] + g * t1[5]);
+   dest[3][1] =  a * t1[2] - b * t1[4] + c * t1[5];
+   dest[3][2] =-(a * t2[2] - b * t2[4] + c * t2[5]);
+   dest[3][3] =  a * t3[2] - b * t3[4] + c * t3[5]; */
+  v3 = _mm_add_ps(_mm_mul_ps(x2, t5),
+                  _mm_sub_ps(_mm_mul_ps(x0, t2),
+                             _mm_mul_ps(x1, t4)));
+  v3 = _mm_xor_ps(v3, _mm_set_ps(0.f, -0.f, 0.f, -0.f));
+
+  /* determinant */
+  x0 = _mm_shuffle_ps(v0, v1, _MM_SHUFFLE(0, 0, 0, 0));
+  x1 = _mm_shuffle_ps(v2, v3, _MM_SHUFFLE(0, 0, 0, 0));
+  x0 = _mm_shuffle_ps(x0, x1, _MM_SHUFFLE(2, 0, 2, 0));
+
+  x0 = _mm_mul_ps(x0, r0);
+  x0 = _mm_add_ps(x0, glmm_shuff1(x0, 0, 1, 2, 3));
+  x0 = _mm_add_ps(x0, glmm_shuff1(x0, 1, 0, 0, 1));
+  x0 = _mm_rcp_ps(x0);
+
+  glmm_store(dest[0], _mm_mul_ps(v0, x0));
+  glmm_store(dest[1], _mm_mul_ps(v1, x0));
+  glmm_store(dest[2], _mm_mul_ps(v2, x0));
+  glmm_store(dest[3], _mm_mul_ps(v3, x0));
+}
+
 CGLM_INLINE
 void
 glm_mat4_inv_sse2(mat4 mat, mat4 dest) {
@@ -164,14 +286,14 @@ glm_mat4_inv_sse2(mat4 mat, mat4 dest) {
          x0, x1, x2, x3, x4, x5, x6, x7;
 
   /* 127 <- 0 */
-  r0 = _mm_load_ps(mat[0]); /* d c b a */
-  r1 = _mm_load_ps(mat[1]); /* h g f e */
-  r2 = _mm_load_ps(mat[2]); /* l k j i */
-  r3 = _mm_load_ps(mat[3]); /* p o n m */
+  r0 = glmm_load(mat[0]); /* d c b a */
+  r1 = glmm_load(mat[1]); /* h g f e */
+  r2 = glmm_load(mat[2]); /* l k j i */
+  r3 = glmm_load(mat[3]); /* p o n m */
 
   x0 = _mm_shuffle_ps(r2, r3, _MM_SHUFFLE(3, 2, 3, 2));  /* p o l k */
-  x1 = _mm_shuffle1_ps(x0, 1, 3, 3, 3);                  /* l p p p */
-  x2 = _mm_shuffle1_ps(x0, 0, 2, 2, 2);                  /* k o o o */
+  x1 = glmm_shuff1(x0, 1, 3, 3, 3);                      /* l p p p */
+  x2 = glmm_shuff1(x0, 0, 2, 2, 2);                      /* k o o o */
   x0 = _mm_shuffle_ps(r2, r1, _MM_SHUFFLE(3, 3, 3, 3));  /* h h l l */
   x3 = _mm_shuffle_ps(r2, r1, _MM_SHUFFLE(2, 2, 2, 2));  /* g g k k */
 
@@ -182,7 +304,7 @@ glm_mat4_inv_sse2(mat4 mat, mat4 dest) {
   t0 = _mm_sub_ps(_mm_mul_ps(x3, x1), _mm_mul_ps(x2, x0));
 
   x4 = _mm_shuffle_ps(r2, r3, _MM_SHUFFLE(2, 1, 2, 1)); /* o n k j */
-  x4 = _mm_shuffle1_ps(x4, 0, 2, 2, 2);                 /* j n n n */
+  x4 = glmm_shuff1(x4, 0, 2, 2, 2);                     /* j n n n */
   x5 = _mm_shuffle_ps(r2, r1, _MM_SHUFFLE(1, 1, 1, 1)); /* f f j j */
 
   /* t1[1] = j * p - n * l;
@@ -198,7 +320,7 @@ glm_mat4_inv_sse2(mat4 mat, mat4 dest) {
   t2 = _mm_sub_ps(_mm_mul_ps(x5, x2), _mm_mul_ps(x4, x3));
 
   x6 = _mm_shuffle_ps(r2, r1, _MM_SHUFFLE(0, 0, 0, 0)); /* e e i i */
-  x7 = _mm_shuffle2_ps(r3, r2, 0, 0, 0, 0, 2, 0, 0, 0); /* i m m m */
+  x7 = glmm_shuff2(r3, r2, 0, 0, 0, 0, 2, 0, 0, 0);     /* i m m m */
 
   /* t1[3] = i * p - m * l;
      t1[3] = i * p - m * l;
@@ -218,10 +340,10 @@ glm_mat4_inv_sse2(mat4 mat, mat4 dest) {
      t3[5] = e * j - i * f; */
   t5 = _mm_sub_ps(_mm_mul_ps(x6, x4), _mm_mul_ps(x7, x5));
 
-  x0 = _mm_shuffle2_ps(r1, r0, 0, 0, 0, 0, 2, 2, 2, 0); /* a a a e */
-  x1 = _mm_shuffle2_ps(r1, r0, 1, 1, 1, 1, 2, 2, 2, 0); /* b b b f */
-  x2 = _mm_shuffle2_ps(r1, r0, 2, 2, 2, 2, 2, 2, 2, 0); /* c c c g */
-  x3 = _mm_shuffle2_ps(r1, r0, 3, 3, 3, 3, 2, 2, 2, 0); /* d d d h */
+  x0 = glmm_shuff2(r1, r0, 0, 0, 0, 0, 2, 2, 2, 0); /* a a a e */
+  x1 = glmm_shuff2(r1, r0, 1, 1, 1, 1, 2, 2, 2, 0); /* b b b f */
+  x2 = glmm_shuff2(r1, r0, 2, 2, 2, 2, 2, 2, 2, 0); /* c c c g */
+  x3 = glmm_shuff2(r1, r0, 3, 3, 3, 3, 2, 2, 2, 0); /* d d d h */
 
   /*
    dest[0][0] =  f * t1[0] - g * t1[1] + h * t1[2];
@@ -269,138 +391,14 @@ glm_mat4_inv_sse2(mat4 mat, mat4 dest) {
   x0 = _mm_shuffle_ps(x0, x1, _MM_SHUFFLE(2, 0, 2, 0));
 
   x0 = _mm_mul_ps(x0, r0);
-  x0 = _mm_add_ps(x0, _mm_shuffle1_ps(x0, 0, 1, 2, 3));
-  x0 = _mm_add_ps(x0, _mm_shuffle1_ps(x0, 1, 0, 0, 1));
-  x0 = _mm_rcp_ps(x0);
-
-  _mm_store_ps(dest[0], _mm_mul_ps(v0, x0));
-  _mm_store_ps(dest[1], _mm_mul_ps(v1, x0));
-  _mm_store_ps(dest[2], _mm_mul_ps(v2, x0));
-  _mm_store_ps(dest[3], _mm_mul_ps(v3, x0));
-}
-
-CGLM_INLINE
-void
-glm_mat4_inv_precise_sse2(mat4 mat, mat4 dest) {
-  __m128 r0, r1, r2, r3,
-         v0, v1, v2, v3,
-         t0, t1, t2, t3, t4, t5,
-         x0, x1, x2, x3, x4, x5, x6, x7;
-
-  /* 127 <- 0 */
-  r0 = _mm_load_ps(mat[0]); /* d c b a */
-  r1 = _mm_load_ps(mat[1]); /* h g f e */
-  r2 = _mm_load_ps(mat[2]); /* l k j i */
-  r3 = _mm_load_ps(mat[3]); /* p o n m */
-
-  x0 = _mm_shuffle_ps(r2, r3, _MM_SHUFFLE(3, 2, 3, 2));  /* p o l k */
-  x1 = _mm_shuffle1_ps(x0, 1, 3, 3, 3);                  /* l p p p */
-  x2 = _mm_shuffle1_ps(x0, 0, 2, 2, 2);                  /* k o o o */
-  x0 = _mm_shuffle_ps(r2, r1, _MM_SHUFFLE(3, 3, 3, 3));  /* h h l l */
-  x3 = _mm_shuffle_ps(r2, r1, _MM_SHUFFLE(2, 2, 2, 2));  /* g g k k */
-
-  /* t1[0] = k * p - o * l;
-     t1[0] = k * p - o * l;
-     t2[0] = g * p - o * h;
-     t3[0] = g * l - k * h; */
-  t0 = _mm_sub_ps(_mm_mul_ps(x3, x1), _mm_mul_ps(x2, x0));
-
-  x4 = _mm_shuffle_ps(r2, r3, _MM_SHUFFLE(2, 1, 2, 1)); /* o n k j */
-  x4 = _mm_shuffle1_ps(x4, 0, 2, 2, 2);                 /* j n n n */
-  x5 = _mm_shuffle_ps(r2, r1, _MM_SHUFFLE(1, 1, 1, 1)); /* f f j j */
-
-  /* t1[1] = j * p - n * l;
-     t1[1] = j * p - n * l;
-     t2[1] = f * p - n * h;
-     t3[1] = f * l - j * h; */
-  t1 = _mm_sub_ps(_mm_mul_ps(x5, x1), _mm_mul_ps(x4, x0));
-
-  /* t1[2] = j * o - n * k
-     t1[2] = j * o - n * k;
-     t2[2] = f * o - n * g;
-     t3[2] = f * k - j * g; */
-  t2 = _mm_sub_ps(_mm_mul_ps(x5, x2), _mm_mul_ps(x4, x3));
-
-  x6 = _mm_shuffle_ps(r2, r1, _MM_SHUFFLE(0, 0, 0, 0)); /* e e i i */
-  x7 = _mm_shuffle2_ps(r3, r2, 0, 0, 0, 0, 2, 0, 0, 0); /* i m m m */
-
-  /* t1[3] = i * p - m * l;
-     t1[3] = i * p - m * l;
-     t2[3] = e * p - m * h;
-     t3[3] = e * l - i * h; */
-  t3 = _mm_sub_ps(_mm_mul_ps(x6, x1), _mm_mul_ps(x7, x0));
-
-  /* t1[4] = i * o - m * k;
-     t1[4] = i * o - m * k;
-     t2[4] = e * o - m * g;
-     t3[4] = e * k - i * g; */
-  t4 = _mm_sub_ps(_mm_mul_ps(x6, x2), _mm_mul_ps(x7, x3));
-
-  /* t1[5] = i * n - m * j;
-     t1[5] = i * n - m * j;
-     t2[5] = e * n - m * f;
-     t3[5] = e * j - i * f; */
-  t5 = _mm_sub_ps(_mm_mul_ps(x6, x4), _mm_mul_ps(x7, x5));
-
-  x0 = _mm_shuffle2_ps(r1, r0, 0, 0, 0, 0, 2, 2, 2, 0); /* a a a e */
-  x1 = _mm_shuffle2_ps(r1, r0, 1, 1, 1, 1, 2, 2, 2, 0); /* b b b f */
-  x2 = _mm_shuffle2_ps(r1, r0, 2, 2, 2, 2, 2, 2, 2, 0); /* c c c g */
-  x3 = _mm_shuffle2_ps(r1, r0, 3, 3, 3, 3, 2, 2, 2, 0); /* d d d h */
-
-  /*
-   dest[0][0] =  f * t1[0] - g * t1[1] + h * t1[2];
-   dest[0][1] =-(b * t1[0] - c * t1[1] + d * t1[2]);
-   dest[0][2] =  b * t2[0] - c * t2[1] + d * t2[2];
-   dest[0][3] =-(b * t3[0] - c * t3[1] + d * t3[2]); */
-  v0 = _mm_add_ps(_mm_mul_ps(x3, t2),
-                  _mm_sub_ps(_mm_mul_ps(x1, t0),
-                             _mm_mul_ps(x2, t1)));
-  v0 = _mm_xor_ps(v0, _mm_set_ps(-0.f, 0.f, -0.f, 0.f));
-
-  /*
-   dest[1][0] =-(e * t1[0] - g * t1[3] + h * t1[4]);
-   dest[1][1] =  a * t1[0] - c * t1[3] + d * t1[4];
-   dest[1][2] =-(a * t2[0] - c * t2[3] + d * t2[4]);
-   dest[1][3] =  a * t3[0] - c * t3[3] + d * t3[4]; */
-  v1 = _mm_add_ps(_mm_mul_ps(x3, t4),
-                  _mm_sub_ps(_mm_mul_ps(x0, t0),
-                             _mm_mul_ps(x2, t3)));
-  v1 = _mm_xor_ps(v1, _mm_set_ps(0.f, -0.f, 0.f, -0.f));
-
-  /*
-   dest[2][0] =  e * t1[1] - f * t1[3] + h * t1[5];
-   dest[2][1] =-(a * t1[1] - b * t1[3] + d * t1[5]);
-   dest[2][2] =  a * t2[1] - b * t2[3] + d * t2[5];
-   dest[2][3] =-(a * t3[1] - b * t3[3] + d * t3[5]);*/
-  v2 = _mm_add_ps(_mm_mul_ps(x3, t5),
-                  _mm_sub_ps(_mm_mul_ps(x0, t1),
-                             _mm_mul_ps(x1, t3)));
-  v2 = _mm_xor_ps(v2, _mm_set_ps(-0.f, 0.f, -0.f, 0.f));
-
-  /*
-   dest[3][0] =-(e * t1[2] - f * t1[4] + g * t1[5]);
-   dest[3][1] =  a * t1[2] - b * t1[4] + c * t1[5];
-   dest[3][2] =-(a * t2[2] - b * t2[4] + c * t2[5]);
-   dest[3][3] =  a * t3[2] - b * t3[4] + c * t3[5]; */
-  v3 = _mm_add_ps(_mm_mul_ps(x2, t5),
-                  _mm_sub_ps(_mm_mul_ps(x0, t2),
-                             _mm_mul_ps(x1, t4)));
-  v3 = _mm_xor_ps(v3, _mm_set_ps(0.f, -0.f, 0.f, -0.f));
-
-  /* determinant */
-  x0 = _mm_shuffle_ps(v0, v1, _MM_SHUFFLE(0, 0, 0, 0));
-  x1 = _mm_shuffle_ps(v2, v3, _MM_SHUFFLE(0, 0, 0, 0));
-  x0 = _mm_shuffle_ps(x0, x1, _MM_SHUFFLE(2, 0, 2, 0));
-
-  x0 = _mm_mul_ps(x0, r0);
-  x0 = _mm_add_ps(x0, _mm_shuffle1_ps(x0, 0, 1, 2, 3));
-  x0 = _mm_add_ps(x0, _mm_shuffle1_ps(x0, 1, 0, 0, 1));
+  x0 = _mm_add_ps(x0, glmm_shuff1(x0, 0, 1, 2, 3));
+  x0 = _mm_add_ps(x0, glmm_shuff1(x0, 1, 0, 0, 1));
   x0 = _mm_div_ps(_mm_set1_ps(1.0f), x0);
 
-  _mm_store_ps(dest[0], _mm_mul_ps(v0, x0));
-  _mm_store_ps(dest[1], _mm_mul_ps(v1, x0));
-  _mm_store_ps(dest[2], _mm_mul_ps(v2, x0));
-  _mm_store_ps(dest[3], _mm_mul_ps(v3, x0));
+  glmm_store(dest[0], _mm_mul_ps(v0, x0));
+  glmm_store(dest[1], _mm_mul_ps(v1, x0));
+  glmm_store(dest[2], _mm_mul_ps(v2, x0));
+  glmm_store(dest[3], _mm_mul_ps(v3, x0));
 }
 
 #endif

include/cglm/simd/sse2/quat.h

@@ -14,56 +14,33 @@
 
 CGLM_INLINE
 void
-glm_quat_slerp_sse2(versor q,
-                    versor r,
-                    float  t,
-                    versor dest) {
-  /* https://en.wikipedia.org/wiki/Slerp */
-  float cosTheta, sinTheta, angle, a, b, c;
+glm_quat_mul_sse2(versor p, versor q, versor dest) {
+  /*
+   + (a1 b2 + b1 a2 + c1 d2 − d1 c2)i
+   + (a1 c2 − b1 d2 + c1 a2 + d1 b2)j
+   + (a1 d2 + b1 c2 − c1 b2 + d1 a2)k
+     a1 a2 − b1 b2 − c1 c2 − d1 d2
+   */
 
-  __m128 xmm_q;
+  __m128 xp, xq, x0, r;
 
-  xmm_q = _mm_load_ps(q);
+  xp = glmm_load(p); /* 3 2 1 0 */
+  xq = glmm_load(q);
 
-  cosTheta = glm_vec4_dot(q, r);
-  if (cosTheta < 0.0f) {
-    _mm_store_ps(q,
-                 _mm_xor_ps(xmm_q,
-                            _mm_set1_ps(-0.f))) ;
+  r  = _mm_mul_ps(glmm_shuff1x(xp, 3), xq);
 
-    cosTheta = -cosTheta;
-  }
+  x0 = _mm_xor_ps(glmm_shuff1x(xp, 0), _mm_set_ps(-0.f, 0.f, -0.f, 0.f));
+  r  = _mm_add_ps(r, _mm_mul_ps(x0, glmm_shuff1(xq, 0, 1, 2, 3)));
 
-  if (cosTheta >= 1.0f) {
-    _mm_store_ps(dest, xmm_q);
-    return;
-  }
+  x0 = _mm_xor_ps(glmm_shuff1x(xp, 1), _mm_set_ps(-0.f, -0.f, 0.f, 0.f));
+  r  = _mm_add_ps(r, _mm_mul_ps(x0, glmm_shuff1(xq, 1, 0, 3, 2)));
 
-  sinTheta = sqrtf(1.0f - cosTheta * cosTheta);
+  x0 = _mm_xor_ps(glmm_shuff1x(xp, 2), _mm_set_ps(-0.f, 0.f, 0.f, -0.f));
+  r  = _mm_add_ps(r, _mm_mul_ps(x0, glmm_shuff1(xq, 2, 3, 0, 1)));
 
-  c = 1.0f - t;
-
-  /* LERP */
-  if (sinTheta < 0.001f) {
-    _mm_store_ps(dest, _mm_add_ps(_mm_mul_ps(_mm_set1_ps(c),
-                                             xmm_q),
-                                  _mm_mul_ps(_mm_set1_ps(t),
-                                             _mm_load_ps(r))));
-    return;
-  }
-
-  /* SLERP */
-  angle = acosf(cosTheta);
-  a = sinf(c * angle);
-  b = sinf(t * angle);
-
-  _mm_store_ps(dest,
-               _mm_div_ps(_mm_add_ps(_mm_mul_ps(_mm_set1_ps(a),
-                                                xmm_q),
-                                     _mm_mul_ps(_mm_set1_ps(b),
-                                                _mm_load_ps(r))),
-                          _mm_set1_ps(sinTheta)));
+  glmm_store(dest, r);
 }
 
+
 #endif
 #endif /* cglm_quat_simd_h */

include/cglm/simd/x86.h

@@ -0,0 +1,192 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglm_simd_x86_h
+#define cglm_simd_x86_h
+#include "intrin.h"
+#ifdef CGLM_SIMD_x86
+
+#ifdef CGLM_ALL_UNALIGNED
+#  define glmm_load(p)      _mm_loadu_ps(p)
+#  define glmm_store(p, a)  _mm_storeu_ps(p, a)
+#else
+#  define glmm_load(p)      _mm_load_ps(p)
+#  define glmm_store(p, a)  _mm_store_ps(p, a)
+#endif
+
+#ifdef CGLM_USE_INT_DOMAIN
+#  define glmm_shuff1(xmm, z, y, x, w)                                        \
+     _mm_castsi128_ps(_mm_shuffle_epi32(_mm_castps_si128(xmm),                \
+                                        _MM_SHUFFLE(z, y, x, w)))
+#else
+#  define glmm_shuff1(xmm, z, y, x, w)                                        \
+       _mm_shuffle_ps(xmm, xmm, _MM_SHUFFLE(z, y, x, w))
+#endif
+
+#define glmm_shuff1x(xmm, x) glmm_shuff1(xmm, x, x, x, x)
+#define glmm_shuff2(a, b, z0, y0, x0, w0, z1, y1, x1, w1)                     \
+     glmm_shuff1(_mm_shuffle_ps(a, b, _MM_SHUFFLE(z0, y0, x0, w0)),           \
+                 z1, y1, x1, w1)
+
+#ifdef __AVX__
+#  ifdef CGLM_ALL_UNALIGNED
+#    define glmm_load256(p)      _mm256_loadu_ps(p)
+#    define glmm_store256(p, a)  _mm256_storeu_ps(p, a)
+#  else
+#    define glmm_load256(p)      _mm256_load_ps(p)
+#    define glmm_store256(p, a)  _mm256_store_ps(p, a)
+#  endif
+#endif
+
+static inline
+__m128
+glmm_abs(__m128 x) {
+  return _mm_andnot_ps(_mm_set1_ps(-0.0f), x);
+}
+
+static inline
+__m128
+glmm_vhadds(__m128 v) {
+#if defined(__SSE3__)
+  __m128 shuf, sums;
+  shuf = _mm_movehdup_ps(v);
+  sums = _mm_add_ps(v, shuf);
+  shuf = _mm_movehl_ps(shuf, sums);
+  sums = _mm_add_ss(sums, shuf);
+  return sums;
+#else
+  __m128 shuf, sums;
+  shuf = glmm_shuff1(v, 2, 3, 0, 1);
+  sums = _mm_add_ps(v, shuf);
+  shuf = _mm_movehl_ps(shuf, sums);
+  sums = _mm_add_ss(sums, shuf);
+  return sums;
+#endif
+}
+
+static inline
+float
+glmm_hadd(__m128 v) {
+  return _mm_cvtss_f32(glmm_vhadds(v));
+}
+
+static inline
+__m128
+glmm_vhmin(__m128 v) {
+  __m128 x0, x1, x2;
+  x0 = _mm_movehl_ps(v, v);     /* [2, 3, 2, 3] */
+  x1 = _mm_min_ps(x0, v);       /* [0|2, 1|3, 2|2, 3|3] */
+  x2 = glmm_shuff1x(x1, 1);     /* [1|3, 1|3, 1|3, 1|3] */
+  return _mm_min_ss(x1, x2);
+}
+
+static inline
+float
+glmm_hmin(__m128 v) {
+  return _mm_cvtss_f32(glmm_vhmin(v));
+}
+
+static inline
+__m128
+glmm_vhmax(__m128 v) {
+  __m128 x0, x1, x2;
+  x0 = _mm_movehl_ps(v, v);     /* [2, 3, 2, 3] */
+  x1 = _mm_max_ps(x0, v);       /* [0|2, 1|3, 2|2, 3|3] */
+  x2 = glmm_shuff1x(x1, 1);     /* [1|3, 1|3, 1|3, 1|3] */
+  return _mm_max_ss(x1, x2);
+}
+
+static inline
+float
+glmm_hmax(__m128 v) {
+  return _mm_cvtss_f32(glmm_vhmax(v));
+}
+
+static inline
+__m128
+glmm_vdots(__m128 a, __m128 b) {
+#if (defined(__SSE4_1__) || defined(__SSE4_2__)) && defined(CGLM_SSE4_DOT)
+  return _mm_dp_ps(a, b, 0xFF);
+#elif defined(__SSE3__) && defined(CGLM_SSE3_DOT)
+  __m128 x0, x1;
+  x0 = _mm_mul_ps(a, b);
+  x1 = _mm_hadd_ps(x0, x0);
+  return _mm_hadd_ps(x1, x1);
+#else
+  return glmm_vhadds(_mm_mul_ps(a, b));
+#endif
+}
+
+static inline
+__m128
+glmm_vdot(__m128 a, __m128 b) {
+#if (defined(__SSE4_1__) || defined(__SSE4_2__)) && defined(CGLM_SSE4_DOT)
+  return _mm_dp_ps(a, b, 0xFF);
+#elif defined(__SSE3__) && defined(CGLM_SSE3_DOT)
+  __m128 x0, x1;
+  x0 = _mm_mul_ps(a, b);
+  x1 = _mm_hadd_ps(x0, x0);
+  return _mm_hadd_ps(x1, x1);
+#else
+  __m128 x0;
+  x0 = _mm_mul_ps(a, b);
+  x0 = _mm_add_ps(x0, glmm_shuff1(x0, 1, 0, 3, 2));
+  return _mm_add_ps(x0, glmm_shuff1(x0, 0, 1, 0, 1));
+#endif
+}
+
+static inline
+float
+glmm_dot(__m128 a, __m128 b) {
+  return _mm_cvtss_f32(glmm_vdots(a, b));
+}
+
+static inline
+float
+glmm_norm(__m128 a) {
+  return _mm_cvtss_f32(_mm_sqrt_ss(glmm_vhadds(_mm_mul_ps(a, a))));
+}
+
+static inline
+float
+glmm_norm2(__m128 a) {
+  return _mm_cvtss_f32(glmm_vhadds(_mm_mul_ps(a, a)));
+}
+
+static inline
+float
+glmm_norm_one(__m128 a) {
+  return _mm_cvtss_f32(glmm_vhadds(glmm_abs(a)));
+}
+
+static inline
+float
+glmm_norm_inf(__m128 a) {
+  return _mm_cvtss_f32(glmm_vhmax(glmm_abs(a)));
+}
+
+static inline
+__m128
+glmm_load3(float v[3]) {
+  __m128i xy;
+  __m128  z;
+
+  xy = _mm_loadl_epi64((const __m128i *)v);
+  z  = _mm_load_ss(&v[2]);
+
+  return _mm_movelh_ps(_mm_castsi128_ps(xy), z);
+}
+
+static inline
+void
+glmm_store3(float v[3], __m128 vx) {
+  _mm_storel_pi((__m64 *)&v[0], vx);
+  _mm_store_ss(&v[2], glmm_shuff1(vx, 2, 2, 2, 2));
+}
+
+#endif
+#endif /* cglm_simd_x86_h */

include/cglm/sphere.h

@@ -0,0 +1,99 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglm_sphere_h
+#define cglm_sphere_h
+
+#include "common.h"
+#include "mat4.h"
+
+/*
+  Sphere Representation in cglm: [center.x, center.y, center.z, radii]
+
+  You could use this representation or you can convert it to vec4 before call
+  any function
+ */
+
+/*!
+ * @brief helper for getting sphere radius
+ *
+ * @param[in]   s  sphere
+ *
+ * @return returns radii
+ */
+CGLM_INLINE
+float
+glm_sphere_radii(vec4 s) {
+  return s[3];
+}
+
+/*!
+ * @brief apply transform to sphere, it is just wrapper for glm_mat4_mulv3
+ *
+ * @param[in]  s    sphere
+ * @param[in]  m    transform matrix
+ * @param[out] dest transformed sphere
+ */
+CGLM_INLINE
+void
+glm_sphere_transform(vec4 s, mat4 m, vec4 dest) {
+  glm_mat4_mulv3(m, s, 1.0f, dest);
+  dest[3] = s[3];
+}
+
+/*!
+ * @brief merges two spheres and creates a new one
+ *
+ * two sphere must be in same space, for instance if one in world space then
+ * the other must be in world space too, not in local space.
+ *
+ * @param[in]  s1   sphere 1
+ * @param[in]  s2   sphere 2
+ * @param[out] dest merged/extended sphere
+ */
+CGLM_INLINE
+void
+glm_sphere_merge(vec4 s1, vec4 s2, vec4 dest) {
+  float dist, radii;
+
+  dist  = glm_vec3_distance(s1, s2);
+  radii = dist + s1[3] + s2[3];
+
+  radii = glm_max(radii, s1[3]);
+  radii = glm_max(radii, s2[3]);
+
+  glm_vec3_center(s1, s2, dest);
+  dest[3] = radii;
+}
+
+/*!
+ * @brief check if two sphere intersects
+ *
+ * @param[in]   s1  sphere
+ * @param[in]   s2  other sphere
+ */
+CGLM_INLINE
+bool
+glm_sphere_sphere(vec4 s1, vec4 s2) {
+  return glm_vec3_distance2(s1, s2) <= glm_pow2(s1[3] + s2[3]);
+}
+
+/*!
+ * @brief check if sphere intersects with point
+ *
+ * @param[in]   s      sphere
+ * @param[in]   point  point
+ */
+CGLM_INLINE
+bool
+glm_sphere_point(vec4 s, vec3 point) {
+  float rr;
+  rr = s[3] * s[3];
+  return glm_vec3_distance2(point, s) <= rr;
+}
+
+#endif /* cglm_sphere_h */

include/cglm/struct.h

@@ -0,0 +1,38 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglm_structs_h
+#define cglm_structs_h
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "cglm.h"
+#include "types-struct.h"
+#include "struct/vec2.h"
+#include "struct/vec3.h"
+#include "struct/vec4.h"
+#include "struct/mat2.h"
+#include "struct/mat3.h"
+#include "struct/mat4.h"
+#include "struct/affine.h"
+#include "struct/frustum.h"
+#include "struct/plane.h"
+#include "struct/box.h"
+#include "struct/color.h"
+#include "struct/io.h"
+#include "struct/cam.h"
+#include "struct/quat.h"
+#include "struct/euler.h"
+#include "struct/project.h"
+#include "struct/sphere.h"
+#include "struct/curve.h"
+
+#ifdef __cplusplus
+}
+#endif
+#endif /* cglm_structs_h */