Merge pull request #88 from yushli/master

Update quat.h

.travis.yml

@@ -57,3 +57,6 @@ after_success:
         --gcov-options '\-lp'
         --verbose;
     fi
+
+after_failure:
+  - cat ./test-suite.log

configure.ac

@@ -7,7 +7,7 @@
 #*****************************************************************************
 
 AC_PREREQ([2.69])
-AC_INIT([cglm], [0.5.2], [info@recp.me])
+AC_INIT([cglm], [0.5.4], [info@recp.me])
 AM_INIT_AUTOMAKE([-Wall -Werror foreign subdir-objects])
 
 AC_CONFIG_MACRO_DIR([m4])

docs/source/build.rst

@@ -1,9 +1,7 @@
-Building cglm
+Build cglm
 ================================
 
-| **cglm** does not have external dependencies except for unit testing.
+| **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.
-| 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
@@ -12,7 +10,8 @@ External dependencies:
 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:
@@ -26,11 +25,12 @@ 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 +50,19 @@ 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/conf.py

@@ -62,9 +62,9 @@ author = u'Recep Aslantas'
 # built documents.
 #
 # The short X.Y version.
-version = u'0.5.2'
+version = u'0.5.4'
 # The full version, including alpha/beta/rc tags.
-release = u'0.5.2'
+release = u'0.5.4'
 
 # The language for content autogenerated by Sphinx. Refer to documentation
 # for a list of supported languages.
@@ -90,7 +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
@@ -99,13 +99,13 @@ html_theme = 'alabaster'
 # html_theme_options = {}
 
 html_theme_options = {
-    'github_banner': 'true',
+    # 'github_banner': 'true',
-    'github_button': 'true',
+    # 'github_button': 'true',
-    'github_user': 'recp',
+    # 'github_user': 'recp',
-    'github_repo': 'cglm',
+    # 'github_repo': 'cglm',
-    'travis_button': 'true',
+    # 'travis_button': 'true',
-    'show_related': 'true',
+    # 'show_related': 'true',
-    'fixed_sidebar': 'true'
+    # 'fixed_sidebar': 'true'
 }
 
 # Add any paths that contain custom static files (such as style sheets) here,

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/getting_started.rst

@@ -9,23 +9,26 @@ Types:
 .. code-block:: c
   :linenos:
 
-   typedef float vec3[3];
+  typedef float                   vec2[2];
-   typedef int  ivec3[3];
+  typedef float                   vec3[3];
-   typedef CGLM_ALIGN(16) float vec4[4];
+  typedef int                    ivec3[3];
+  typedef CGLM_ALIGN_IF(16) float vec4[4];
+  typedef vec4                    versor;
+  typedef vec3                    mat3[3];
 
-   typedef vec3 mat3[3];
+  #ifdef __AVX__
-   typedef vec4 mat4[4];
+  typedef CGLM_ALIGN_IF(32) vec4  mat4[4];
-
+  #else
-   typedef vec4 versor;
+  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*
 
-Alignment is Required:
+Alignment Is Required:
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
-**vec4** and **mat4** requires 16 byte alignment because vec4 and mat4 operations are
+**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).
-vectorized by SIMD instructions (SSE/AVX).
 
 **UPDATE:**
   By starting v0.4.5 cglm provides an option to disable alignment requirement, it is enabled as default
@@ -37,10 +40,9 @@ vectorized by SIMD instructions (SSE/AVX).
 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.
+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.
-When allocating memory don't forget that **vec4** and **mat4** requires alignment.
 
-**NOTE:** Unaligned vec4 and unaligned mat4 operations will be supported in the future. Check todo list.
+**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**
 

docs/source/index.rst

@@ -3,7 +3,7 @@
    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).
@@ -14,33 +14,36 @@ is considered to be supported as optional.
 
 Also currently only **float** type is supported for most operations.
 
-**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...)
-
-
 .. toctree::
-   :maxdepth: 1
+   :maxdepth: 2
-   :caption: Table Of Contents:
+   :caption: Getting Started:
 
+   features
    build
    getting_started
+
+.. toctree::
+   :maxdepth: 2
+   :caption: How To:
+
    opengl
+
+.. toctree::
+   :maxdepth: 2
+   :caption: API:
+
    api
+
+.. toctree::
+   :maxdepth: 2
+   :caption: Options:
+
    opt
+
+.. toctree::
+   :maxdepth: 2
+   :caption: Troubleshooting:
+
    troubleshooting
 
 Indices and tables

docs/source/mat3.rst

@@ -21,6 +21,7 @@ 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`
@@ -32,6 +33,7 @@ Functions:
 #. :c:func:`glm_mat3_trace`
 #. :c:func:`glm_mat3_swap_col`
 #. :c:func:`glm_mat3_swap_row`
+#. :c:func:`glm_mat3_rmc`
 
 Functions documentation
 ~~~~~~~~~~~~~~~~~~~~~~~
@@ -59,6 +61,13 @@ Functions documentation
       | *[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
@@ -161,3 +170,20 @@ Functions documentation
       | *[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

@@ -26,6 +26,7 @@ Functions:
 #. :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`
@@ -81,6 +82,13 @@ Functions documentation
       | *[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
@@ -278,7 +286,7 @@ Functions documentation
 
     | helper for  R (row vector) * M (matrix) * C (column vector)
 
-    | the result is scalar because S * M = Matrix1x4 (row vector),
+    | the result is scalar because R * M = Matrix1x4 (row vector),
     | then Matrix1x4 * Vec4 (column vector) = Matrix1x1 (Scalar)
 
     Parameters:

docs/source/opengl.rst

@@ -43,9 +43,9 @@ array of matrices:
    /* ... */
    glUniformMatrix4fv(location, count, GL_FALSE, (float *)matrix);
 
-in this way, passing aray of matrices is same 
+in this way, passing aray of matrices is same
 
-Passing / Uniforming Vectors to OpenGL:¶
+Passing / Uniforming Vectors to OpenGL:
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 You don't need to do extra thing when passing cglm vectors to OpengL or other APIs.

docs/source/opt.rst

@@ -40,3 +40,13 @@ 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.

include/cglm/affine-mat.h

@@ -152,7 +152,7 @@ glm_inv_tr(mat4 mat) {
   glm_inv_tr_sse2(mat);
 #else
   CGLM_ALIGN_MAT mat3 r;
-  CGLM_ALIGN(16) vec3 t;
+  CGLM_ALIGN(8)  vec3 t;
 
   /* rotate */
   glm_mat4_pick3t(mat, r);

include/cglm/bezier.h

@@ -8,6 +8,8 @@
 #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},                   \

include/cglm/call/mat3.h

@@ -72,6 +72,10 @@ 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/cam.h

@@ -84,7 +84,7 @@ glm_frustum(float left,
             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);
@@ -122,7 +122,7 @@ glm_ortho(float left,
           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);
@@ -259,7 +259,7 @@ 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);

include/cglm/common.h

@@ -11,8 +11,10 @@
 #define _USE_MATH_DEFINES /* for windows */
 
 #include <stdint.h>
+#include <stddef.h>
 #include <math.h>
 #include <float.h>
+#include <stdbool.h>
 
 #if defined(_MSC_VER)
 #  ifdef CGLM_DLL
@@ -26,34 +28,6 @@
 #  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)
-
 #include "types.h"
 #include "simd/intrin.h"
 

include/cglm/mat3.h

@@ -17,16 +17,19 @@
    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
@@ -63,7 +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];
 }
 
 /*!
@@ -106,6 +119,18 @@ glm_mat3_identity_array(mat3 * __restrict mat, size_t count) {
   }
 }
 
+/*!
+ * @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);
+}
+
 /*!
  * @brief multiply m1 and m2 to dest
  *
@@ -372,4 +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

@@ -22,6 +22,7 @@
    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);
@@ -31,6 +32,7 @@
    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);
@@ -40,6 +42,7 @@
    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
@@ -98,7 +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];
 }
 
 /*!
@@ -168,6 +179,18 @@ glm_mat4_identity_array(mat4 * __restrict mat, size_t count) {
   }
 }
 
+/*!
+ * @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);
+}
+
 /*!
  * @brief copy upper-left of mat4 to mat3
  *
@@ -474,10 +497,8 @@ glm_mat4_transpose(mat4 m) {
   glm_mat4_transp_sse2(m, m);
 #else
   mat4 d;
-
   glm_mat4_transpose_to(m, d);
-
+  glm_mat4_ucopy(d, m);
-  glm__memcpy(float, m, d, sizeof(mat4));
 #endif
 }
 
@@ -682,7 +703,7 @@ glm_mat4_swap_row(mat4 mat, int row1, int row2) {
  *
  * rmc stands for Row * Matrix * Column
  *
- * the result is scalar because S * M = Matrix1x4 (row vector),
+ * the result is scalar because R * M = Matrix1x4 (row vector),
  * then Matrix1x4 * Vec4 (column vector) = Matrix1x1 (Scalar)
  *
  * @param[in]  r   row vector or matrix1x4

include/cglm/project.h

@@ -8,6 +8,7 @@
 #ifndef cglm_project_h
 #define cglm_project_h
 
+#include "common.h"
 #include "vec3.h"
 #include "vec4.h"
 #include "mat4.h"

include/cglm/quat.h

@@ -233,7 +233,7 @@ glm_quat_normalize_to(versor q, versor dest) {
   dot = glm_vec4_norm2(q);
 
   if (dot <= 0.0f) {
-    glm_quat_identity(q);
+    glm_quat_identity(dest);
     return;
   }
 

include/cglm/util.h

@@ -19,7 +19,6 @@
 #define cglm_util_h
 
 #include "common.h"
-#include <stdbool.h>
 
 #define GLM_MIN(X, Y) (((X) < (Y)) ? (X) : (Y))
 #define GLM_MAX(X, Y) (((X) > (Y)) ? (X) : (Y))

include/cglm/vec3-ext.h

@@ -31,9 +31,6 @@
 
 #include "common.h"
 #include "util.h"
-#include <stdbool.h>
-#include <math.h>
-#include <float.h>
 
 /*!
  * @brief fill a vector with specified value

include/cglm/vec3.h

@@ -428,8 +428,8 @@ glm_vec3_maxadd(vec3 a, vec3 b, vec3 dest) {
  *
  * it applies += operator so dest must be initialized
  *
- * @param[in]  a    vector
+ * @param[in]  a    vector 1
- * @param[in]  b    scalar
+ * @param[in]  b    vector 2
  * @param[out] dest dest += min(a, b)
  */
 CGLM_INLINE

include/cglm/vec4-ext.h

@@ -31,9 +31,6 @@
 
 #include "common.h"
 #include "vec3-ext.h"
-#include <stdbool.h>
-#include <math.h>
-#include <float.h>
 
 /*!
  * @brief fill a vector with specified value

include/cglm/vec4.h

@@ -576,8 +576,8 @@ glm_vec4_maxadd(vec4 a, vec4 b, vec4 dest) {
  *
  * it applies += operator so dest must be initialized
  *
- * @param[in]  a    vector
+ * @param[in]  a    vector 1
- * @param[in]  b    scalar
+ * @param[in]  b    vector 2
  * @param[out] dest dest += min(a, b)
  */
 CGLM_INLINE
@@ -690,18 +690,9 @@ CGLM_INLINE
 float
 glm_vec4_distance(vec4 a, vec4 b) {
 #if defined( __SSE__ ) || defined( __SSE2__ )
-  __m128 x0;
+  return glmm_norm(_mm_sub_ps(glmm_load(b), glmm_load(a)));
-  x0 = _mm_sub_ps(glmm_load(b), glmm_load(a));
-  x0 = _mm_mul_ps(x0, x0);
-  x0 = _mm_add_ps(x0, glmm_shuff1(x0, 1, 0, 3, 2));
-  return _mm_cvtss_f32(_mm_sqrt_ss(_mm_add_ss(x0,
-                                              glmm_shuff1(x0, 0, 1, 0, 1))));
 #elif defined(CGLM_NEON_FP)
-  float32x4_t v0;
+  return glmm_norm(vsubq_f32(glmm_load(a), glmm_load(b)));
-  float32_t   r;
-  v0 = vsubq_f32(vld1q_f32(a), vld1q_f32(b));
-  r  = vaddvq_f32(vmulq_f32(v0, v0));
-  return sqrtf(r);
 #else
   return sqrtf(glm_pow2(b[0] - a[0])
              + glm_pow2(b[1] - a[1])

include/cglm/version.h

@@ -10,6 +10,6 @@
 
 #define CGLM_VERSION_MAJOR 0
 #define CGLM_VERSION_MINOR 5
-#define CGLM_VERSION_PATCH 2
+#define CGLM_VERSION_PATCH 4
 
 #endif /* cglm_version_h */

post-build.sh

@@ -8,12 +8,17 @@
 
 cd $(dirname "$0")
 
-mkdir -p .libs
+mkdir -p "$(pwd)/.libs"
+
+libmocka_folder=$(pwd)/test/lib/cmocka/build/src/
 
 if [ "$(uname)" = "Darwin" ]; then
-  ln -sf $(pwd)/test/lib/cmocka/build/src/libcmocka.0.dylib \
+  libcmocka=libcmocka.0.dylib
-      .libs/libcmocka.0.dylib;
 else
-  ln -sf $(pwd)/test/lib/cmocka/build/src/libcmocka.so.0 \
+  libcmocka=libcmocka.so.0
-      .libs/libcmocka.so.0;
+fi
+
+libcmocka_path="$libmocka_folder$libcmocka"
+if [ -f "$libcmocka_path" ]; then
+  ln -sf "$libcmocka_path" "$(pwd)/.libs/$libcmocka";
 fi

src/mat3.c

@@ -91,3 +91,9 @@ void
 glmc_mat3_swap_row(mat3 mat, int row1, int row2) {
   glm_mat3_swap_row(mat, row1, row2);
 }
+
+CGLM_EXPORT
+float
+glmc_mat3_rmc(vec3 r, mat3 m, vec3 c) {
+  return glm_mat3_rmc(r, m, c);
+}

test/src/test_bezier.c

@@ -51,8 +51,8 @@ test_bezier(void **state) {
   Bs_plain = test_bezier_plain(s, p0, c0, c1, p1);
 
   assert_true(glm_eq(Bs,  Bs_plain));
-  assert_true(glm_eq(smc, Bs_plain));
+  test_assert_eqf(smc, Bs_plain);
-  assert_true(glm_eq(Bs,  smc));
+  test_assert_eqf(Bs,  smc);
 
   /* test cubic hermite */
   smc      = glm_smc(s, GLM_HERMITE_MAT, (vec4){p0, p1, c0, c1});

test/src/test_common.c

@@ -5,6 +5,7 @@
 
 #include "test_common.h"
 #include <stdlib.h>
+#include <math.h>
 
 #define m 4
 #define n 4

test/src/test_vec4.c

@@ -93,6 +93,13 @@ test_vec4(void **state) {
     /* 3. test SIMD norm2 */
     test_rand_vec4(v);
     test_assert_eqf(test_vec4_norm2(v), glm_vec4_norm2(v));
+
+    /* 4. test SSE/SIMD distance */
+    test_rand_vec4(v1);
+    test_rand_vec4(v2);
+    d1 = glm_vec4_distance(v1, v2);
+    d2 = sqrtf(powf(v1[0]-v2[0], 2.0f) + pow(v1[1]-v2[1], 2.0f) + pow(v1[2]-v2[2], 2.0f) + pow(v1[3]-v2[3], 2.0f));
+    assert_true(fabsf(d1 - d2) <= 0.000009);
   }
 
   /* test zero */