arm neon: multiply mat4 with vec4

* this will help to disable colorful output because not all terminals can print colors.

CMakeLists.txt

@@ -1,5 +1,5 @@
 cmake_minimum_required(VERSION 3.8.2)
-project(cglm VERSION 0.7.7 LANGUAGES C)
+project(cglm VERSION 0.7.9 LANGUAGES C)
 
 set(CMAKE_C_STANDARD 11)
 set(CMAKE_C_STANDARD_REQUIRED YES)
@@ -22,7 +22,7 @@ if(CGLM_USE_C99)
 endif()
 
 if(MSVC)
-  add_definitions(-DNDEBUG -D_WINDOWS -D_USRDLL -DCGLM_EXPORTS -DCGLM_DLL)
+  add_definitions(-DNDEBUG -D_WINDOWS -D_USRDLL)
   add_compile_options(/W3 /Ox /Gy /Oi /TC)
   
   # Ref: https://skia.googlesource.com/third_party/sdl/+/refs/heads/master/CMakeLists.txt#225
@@ -75,6 +75,12 @@ add_library(${PROJECT_NAME}
   src/affine2d.c
   )
 
+if(CGLM_SHARED)
+  add_definitions(-DCGLM_EXPORTS)
+else()
+  target_compile_definitions(${PROJECT_NAME} PUBLIC -DCGLM_STATIC)
+endif()
+
 set_target_properties(${PROJECT_NAME} PROPERTIES
                               VERSION ${PROJECT_VERSION} 
                             SOVERSION ${PROJECT_VERSION_MAJOR})

README.md

@@ -25,6 +25,7 @@ you have the latest version
 - **[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
+- **[major change]** by starting v0.7.3, inline print functions are disabled in release/production mode to eliminate print costs (see options in documentation). Print output also improved. You can disable colors if you need  (see documentation)
 
 #### Note for C++ developers:
 If you are not aware of the original GLM library yet, you may also want to look at:
@@ -183,6 +184,35 @@ add_subdirectory(external/cglm/)
 # or you can use find_package to configure cglm
 ```
 
+### Meson (All platforms)
+
+```bash
+$ meson build # [Optional] --default-library=static
+$ cd build
+$ ninja
+$ sudo ninja install # [Optional]
+```
+
+##### Meson options with Defaults:
+
+```meson
+c_std=c11
+buildtype=release
+default_library=shared
+enable_tests=false # to run tests: ninja test
+```
+#### Use with your Meson project
+* Example:
+```meson
+# Clone cglm or create a cglm.wrap under <source_root>/subprojects
+project('name', 'c')
+
+cglm_dep = dependency('cglm', fallback : 'cglm', 'cglm_dep')
+
+executable('exe', 'src/main.c', dependencies : cglm_dep)
+```
+
+
 ### Unix (Autotools)
 
 ```bash

configure.ac

@@ -7,7 +7,7 @@
 #*****************************************************************************
 
 AC_PREREQ([2.69])
-AC_INIT([cglm], [0.7.7], [info@recp.me])
+AC_INIT([cglm], [0.7.9], [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.

docs/source/build.rst

@@ -47,6 +47,41 @@ If you don't want to install **cglm** to your system's folder you can get static
   
   add_subdirectory(external/cglm/)
 
+Meson (All platforms):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. code-block:: bash
+  :linenos:
+
+  $ meson build # [Optional] --default-library=static
+  $ cd build
+  $ ninja
+  $ sudo ninja install # [Optional]
+
+**Meson Options:**
+
+.. code-block:: CMake
+  :linenos:
+
+  c_std=c11
+  buildtype=release
+  default_library=shared
+  enable_tests=false # to run tests: ninja test
+
+
+**Use with your CMake project example**
+
+.. code-block:: CMake
+  :linenos:
+
+  # Clone cglm or create a cglm.wrap under <source_root>/subprojects
+  project('name', 'c')
+  
+  cglm_dep = dependency('cglm', fallback : 'cglm', 'cglm_dep')
+  
+  executable('exe', 'src/main.c', dependencies : cglm_dep)
+
+
 Unix (Autotools):
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 

docs/source/conf.py

@@ -62,9 +62,9 @@ author = u'Recep Aslantas'
 # built documents.
 #
 # The short X.Y version.
-version = u'0.7.7'
+version = u'0.7.9'
 # The full version, including alpha/beta/rc tags.
-release = u'0.7.7'
+release = u'0.7.9'
 
 # The language for content autogenerated by Sphinx. Refer to documentation
 # for a list of supported languages.

docs/source/opt.rst

@@ -18,7 +18,7 @@ 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
+to define it in build system. If you use pre-compiled versions then you
 have to compile cglm with **CGLM_ALL_UNALIGNED** macro.
 
 **VERY VERY IMPORTANT:** If you use cglm in multiple projects and
@@ -55,11 +55,20 @@ Print Options
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 1. **CGLM_DEFINE_PRINTS**
-2. **CGLM_NO_PRINTS_NOOP**
+2. **CGLM_NO_PRINTS_NOOP** (use CGLM_DEFINE_PRINTS)
 
-Inline prints are only enabled in DEBUG mode and if **CGLM_DEFINE_PRINTS** is defined.
+Inline prints are only enabled in **DEBUG** mode or if **CGLM_DEFINE_PRINTS** is defined.
-If DEBUG is not enabled then print function bodies will be emptied to eliminate print function calls.
+**glmc_** versions will always print too.
-You can disable this feature too by defining **CGLM_NO_PRINTS_NOOP** macro top of cglm headers.
+
+Because **cglm** tried to enable print functions in debug mode and disable them in
+release/production mode to eliminate printing costs when we do not need them.
+
+**cglm** checks **DEBUG** or **_DEBUG** macros to test debug mode, if these are not working for you then you can use 
+**CGLM_DEFINE_PRINTS** to force enable, or create a PR to introduce new macro to test against debugging mode.
+
+If DEBUG mode is not enabled then print functions will be emptied to eliminate print function calls.
+You can disable this feature too by defining **CGLM_DEFINE_PRINTS** macro top of cglm header 
+or in project/build settings...
 
 3. **CGLM_PRINT_PRECISION**    5
 
@@ -71,3 +80,6 @@ if a number is greater than this value then %g will be used, since this is short
 
 5. **CGLM_PRINT_COLOR**        "\033[36m"
 6. **CGLM_PRINT_COLOR_RESET**  "\033[0m"
+
+You can disable colorful print output by defining **CGLM_PRINT_COLOR** and **CGLM_PRINT_COLOR_RESET** as empty macro.
+Because some terminals may not support colors.

include/cglm/common.h

@@ -23,7 +23,9 @@
 #include <stdbool.h>
 
 #if defined(_MSC_VER)
-#  ifdef CGLM_DLL
+#  ifdef CGLM_STATIC
+#    define CGLM_EXPORT
+#  elif defined(CGLM_EXPORTS)
 #    define CGLM_EXPORT __declspec(dllexport)
 #  else
 #    define CGLM_EXPORT __declspec(dllimport)

include/cglm/io.h

@@ -15,20 +15,49 @@
    CGLM_INLINE void glm_versor_print(versor vec, FILE *ostream);
  */
 
+/*
+ cglm tried to enable print functions in debug mode and disable them in
+ release/production mode to eliminate printing costs.
+ 
+ if you need to force enable then define CGLM_DEFINE_PRINTS macro not DEBUG one
+ 
+ Print functions are enabled if:
+ 
+ - DEBUG or _DEBUG macro is defined (mostly defined automatically in debugging)
+ - CGLM_DEFINE_PRINTS macro is defined including release/production
+   which makes enabled printing always
+ - glmc_ calls for io are always prints
+
+ */
+
+/* DEPRECATED: CGLM_NO_PRINTS_NOOP (use CGLM_DEFINE_PRINTS) */
+
 #ifndef cglm_io_h
 #define cglm_io_h
 #if defined(DEBUG) || defined(_DEBUG) \
-   || defined(CGLM_DEFINE_PRINTS) || defined(CGLM_LIB_SRC)
+   || defined(CGLM_DEFINE_PRINTS) || defined(CGLM_LIB_SRC) \
+   || defined(CGLM_NO_PRINTS_NOOP)
 
 #include "common.h"
 
 #include <stdio.h>
 #include <stdlib.h>
 
-#define CGLM_PRINT_PRECISION    5
+#ifndef CGLM_PRINT_PRECISION
-#define CGLM_PRINT_MAX_TO_SHORT 1e5
+#  define CGLM_PRINT_PRECISION    5
-#define CGLM_PRINT_COLOR        "\033[36m"
+#endif
-#define CGLM_PRINT_COLOR_RESET  "\033[0m"
+
+#ifndef CGLM_PRINT_MAX_TO_SHORT
+#  define CGLM_PRINT_MAX_TO_SHORT 1e5
+#endif
+
+#ifndef CGLM_PRINT_COLOR
+#  define CGLM_PRINT_COLOR        "\033[36m"
+#endif
+
+#ifndef CGLM_PRINT_COLOR_RESET
+#  define CGLM_PRINT_COLOR_RESET  "\033[0m"
+#endif
 
 CGLM_INLINE
 void
@@ -293,7 +322,7 @@ glm_aabb_print(vec3                    bbox[2],
 #undef m
 }
 
-#elif !defined(CGLM_NO_PRINTS_NOOP)
+#else
 
 #include "common.h"
 
@@ -301,15 +330,15 @@ glm_aabb_print(vec3                    bbox[2],
 #include <stdlib.h>
 
 /* NOOP: Remove print from DEBUG */
-#define glm_mat4_print(...)
+#define glm_mat4_print(v, s) (void)v; (void)s;
-#define glm_mat3_print(...)
+#define glm_mat3_print(v, s) (void)v; (void)s;
-#define glm_mat2_print(...)
+#define glm_mat2_print(v, s) (void)v; (void)s;
-#define glm_vec4_print(...)
+#define glm_vec4_print(v, s) (void)v; (void)s;
-#define glm_vec3_print(...)
+#define glm_vec3_print(v, s) (void)v; (void)s;
-#define glm_ivec3_print(...)
+#define glm_ivec3_print(v, s) (void)v; (void)s;
-#define glm_vec2_print(...)
+#define glm_vec2_print(v, s) (void)v; (void)s;
-#define glm_versor_print(...)
+#define glm_versor_print(v, s) (void)v; (void)s;
-#define glm_aabb_print(...)
+#define glm_aabb_print(v, t, s) (void)v; (void)t; (void)s;
 
 #endif
 #endif /* cglm_io_h */

include/cglm/mat4.h

@@ -358,6 +358,8 @@ void
 glm_mat4_mulv(mat4 m, vec4 v, vec4 dest) {
 #if defined( __SSE__ ) || defined( __SSE2__ )
   glm_mat4_mulv_sse2(m, v, dest);
+#elif defined(CGLM_NEON_FP)
+  glm_mat4_mulv_neon(m, v, dest);
 #else
   vec4 res;
   res[0] = m[0][0] * v[0] + m[1][0] * v[1] + m[2][0] * v[2] + m[3][0] * v[3];
@@ -476,6 +478,8 @@ void
 glm_mat4_transpose_to(mat4 m, mat4 dest) {
 #if defined( __SSE__ ) || defined( __SSE2__ )
   glm_mat4_transp_sse2(m, dest);
+#elif defined(CGLM_NEON_FP)
+  glm_mat4_transp_neon(m, dest);
 #else
   dest[0][0] = m[0][0]; dest[1][0] = m[0][1];
   dest[0][1] = m[1][0]; dest[1][1] = m[1][1];
@@ -498,6 +502,8 @@ void
 glm_mat4_transpose(mat4 m) {
 #if defined( __SSE__ ) || defined( __SSE2__ )
   glm_mat4_transp_sse2(m, m);
+#elif defined(CGLM_NEON_FP)
+  glm_mat4_transp_neon(m, m);
 #else
   mat4 d;
   glm_mat4_transpose_to(m, d);
@@ -536,12 +542,7 @@ 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;
+  glm_mat4_scale_neon(m, s);
-  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

include/cglm/simd/neon/mat4.h

@@ -12,6 +12,32 @@
 #include "../../common.h"
 #include "../intrin.h"
 
+CGLM_INLINE
+void
+glm_mat4_scale_neon(mat4 m, float s) {
+  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));
+}
+
+CGLM_INLINE
+void
+glm_mat4_transp_neon(mat4 m, mat4 dest) {
+  float32x4x4_t vmat;
+  
+  vmat = vld4q_f32(m[0]);
+
+  vst1q_f32(dest[0], vmat.val[0]);
+  vst1q_f32(dest[1], vmat.val[1]);
+  vst1q_f32(dest[2], vmat.val[2]);
+  vst1q_f32(dest[3], vmat.val[3]);
+}
+
 CGLM_INLINE
 void
 glm_mat4_mul_neon(mat4 m1, mat4 m2, mat4 dest) {
@@ -53,5 +79,27 @@ glm_mat4_mul_neon(mat4 m1, mat4 m2, mat4 dest) {
   vst1q_f32(dest[3], d3);
 }
 
+CGLM_INLINE
+void
+glm_mat4_mulv_neon(mat4 m, vec4 v, vec4 dest) {
+  float32x4_t l0, l1, l2, l3;
+  float32x2_t vlo, vhi;
+  
+  l0  = vld1q_f32(m[0]);
+  l1  = vld1q_f32(m[1]);
+  l2  = vld1q_f32(m[2]);
+  l3  = vld1q_f32(m[3]);
+
+  vlo = vld1_f32(&v[0]);
+  vhi = vld1_f32(&v[2]);
+
+  l0  = vmulq_lane_f32(l0, vlo, 0);
+  l0  = vmlaq_lane_f32(l0, l1, vlo, 1);
+  l0  = vmlaq_lane_f32(l0, l2, vhi, 0);
+  l0  = vmlaq_lane_f32(l0, l3, vhi, 1);
+
+  vst1q_f32(dest, l0);
+}
+
 #endif
 #endif /* cglm_mat4_neon_h */

include/cglm/version.h

@@ -10,6 +10,6 @@
 
 #define CGLM_VERSION_MAJOR 0
 #define CGLM_VERSION_MINOR 7
-#define CGLM_VERSION_PATCH 7
+#define CGLM_VERSION_PATCH 9
 
 #endif /* cglm_version_h */

meson.build

@@ -0,0 +1,107 @@
+project('cglm', 'c',
+    version : '0.7.9', 
+    license : 'mit',
+    default_options : [
+        'c_std=c11',
+        'werror=true',
+        'warning_level=2',
+        'buildtype=release'
+    ]
+)
+
+cc = meson.get_compiler('c')
+
+cglm_deps = cc.find_library('m', required : false)
+
+cglm_args = []
+build_args = []
+
+if get_option('default_library') == 'static'
+    cglm_args = '-DCGLM_STATIC'
+endif
+
+if host_machine.system() == 'windows'
+    build_args = '-DCGLM_EXPORTS'
+endif
+
+cglm_inc = include_directories('include')
+
+cglm_src = files(
+    'src/affine.c',
+    'src/affine2d.c',
+    'src/bezier.c',
+    'src/box.c',
+    'src/cam.c',
+    'src/curve.c',
+    'src/ease.c',
+    'src/euler.c',
+    'src/frustum.c',
+    'src/io.c',
+    'src/mat2.c',
+    'src/mat3.c',
+    'src/mat4.c',
+    'src/plane.c',
+    'src/project.c',
+    'src/quat.c',
+    'src/ray.c',
+    'src/sphere.c',
+    'src/vec2.c',
+    'src/vec3.c',
+    'src/vec4.c'
+)
+
+install_subdir('include/cglm', install_dir : get_option('includedir'))
+
+cglm_lib = library('cglm',
+    cglm_src,
+    install : true,
+    dependencies : cglm_deps,
+    c_args : [ build_args, cglm_args ]
+)
+
+cglm_dep = declare_dependency(
+    link_with : cglm_lib,
+    dependencies : cglm_deps,
+    compile_args : cglm_args,
+    include_directories : cglm_inc,
+    version : meson.project_version()
+)
+
+if meson.version().version_compare('>= 0.54.0')
+    meson.override_dependency('cglm', cglm_dep)
+endif
+
+
+pkg = import('pkgconfig')
+
+pkg.generate(
+    name : 'cglm',
+    libraries : cglm_lib,
+    extra_cflags : cglm_args,
+    version : meson.project_version(),
+    url : 'https://github.com/recp/cglm',
+    description : 'OpenGL Mathematics (glm) for C'
+)
+
+if get_option('build_tests') == true
+
+test_src = files(
+    'test/runner.c',
+    'test/src/test_euler.c',
+    'test/src/test_bezier.c',
+    'test/src/test_cam.c',
+    'test/src/test_struct.c',
+    'test/src/test_clamp.c',
+    'test/src/test_common.c',
+    'test/src/tests.c'
+)
+
+test_exe = executable('tests', 
+    test_src, 
+    dependencies : cglm_dep,
+    c_args : '-DGLM_TESTS_NO_COLORFUL_OUTPUT'
+)
+
+test('cglm.tests', test_exe)
+
+endif

meson_options.txt

@@ -0,0 +1 @@
+option('build_tests', type : 'boolean', value : false, description : 'Build tests')

test/CMakeLists.txt

@@ -18,7 +18,11 @@ set(TEST_RUNNER_PARAMS "")
 add_executable(${TEST_MAIN} ${TESTFILES})
 target_compile_definitions(${TEST_MAIN} PRIVATE CGLM_DEFINE_PRINTS=1)
 
-target_link_libraries(${TEST_MAIN} PRIVATE cglm m)
+if(NOT MSVC)
+  target_link_libraries(${TEST_MAIN} PRIVATE m)
+endif()
+
+target_link_libraries(${TEST_MAIN} PRIVATE cglm)
 target_include_directories(${TEST_MAIN} PRIVATE
   ${CMAKE_CURRENT_LIST_DIR}/include
   ${CMAKE_CURRENT_LIST_DIR}/src

test/include/common.h

@@ -16,6 +16,10 @@
 #  define _CRT_SECURE_NO_WARNINGS /* for windows */
 #endif
 
+#ifndef _GNU_SOURCE
+#  define _GNU_SOURCE /* for drand48() */
+#endif
+
 #include <stdlib.h>
 #include <stdio.h>
 #include <string.h>

test/runner.c

@@ -18,6 +18,8 @@ main(int argc, const char * argv[]) {
   test_status_t st;
   int32_t       i, count, passed, failed, maxlen;
   double        start, end, elapsed, total;
+  (void)argc;
+  (void)argv;
 
   passed = failed = maxlen  = 0;
   total  = 0.0;

win/cglm.vcxproj

@@ -209,7 +209,7 @@
       <PrecompiledHeader>NotUsing</PrecompiledHeader>
       <WarningLevel>Level3</WarningLevel>
       <Optimization>MaxSpeed</Optimization>
-      <PreprocessorDefinitions>WIN32;_DEBUG;_WINDOWS;_USRDLL;CGLM_EXPORTS;CGLM_DLL;%(PreprocessorDefinitions);CGLM_DLL</PreprocessorDefinitions>
+      <PreprocessorDefinitions>WIN32;_DEBUG;_WINDOWS;_USRDLL;CGLM_EXPORTS;%(PreprocessorDefinitions)</PreprocessorDefinitions>
       <DebugInformationFormat>None</DebugInformationFormat>
       <BasicRuntimeChecks>Default</BasicRuntimeChecks>
       <PrecompiledHeaderFile />
@@ -225,7 +225,7 @@
     <ClCompile>
       <WarningLevel>Level3</WarningLevel>
       <Optimization>MaxSpeed</Optimization>
-      <PreprocessorDefinitions>_DEBUG;_WINDOWS;_USRDLL;CGLM_EXPORTS;CGLM_DLL;%(PreprocessorDefinitions);CGLM_DLL</PreprocessorDefinitions>
+      <PreprocessorDefinitions>_DEBUG;_WINDOWS;_USRDLL;CGLM_EXPORTS;%(PreprocessorDefinitions)</PreprocessorDefinitions>
       <InlineFunctionExpansion>AnySuitable</InlineFunctionExpansion>
       <IntrinsicFunctions>true</IntrinsicFunctions>
       <WholeProgramOptimization>true</WholeProgramOptimization>
@@ -246,7 +246,7 @@
       <Optimization>MaxSpeed</Optimization>
       <FunctionLevelLinking>true</FunctionLevelLinking>
       <IntrinsicFunctions>true</IntrinsicFunctions>
-      <PreprocessorDefinitions>WIN32;NDEBUG;_WINDOWS;_USRDLL;CGLM_EXPORTS;CGLM_DLL;%(PreprocessorDefinitions);CGLM_DLL</PreprocessorDefinitions>
+      <PreprocessorDefinitions>WIN32;NDEBUG;_WINDOWS;_USRDLL;CGLM_EXPORTS;%(PreprocessorDefinitions)</PreprocessorDefinitions>
       <DebugInformationFormat>None</DebugInformationFormat>
       <PrecompiledHeaderFile />
       <CompileAs>CompileAsC</CompileAs>
@@ -266,7 +266,7 @@
       <Optimization>MaxSpeed</Optimization>
       <FunctionLevelLinking>true</FunctionLevelLinking>
       <IntrinsicFunctions>true</IntrinsicFunctions>
-      <PreprocessorDefinitions>NDEBUG;_WINDOWS;_USRDLL;CGLM_EXPORTS;CGLM_DLL;%(PreprocessorDefinitions);CGLM_DLL</PreprocessorDefinitions>
+      <PreprocessorDefinitions>NDEBUG;_WINDOWS;_USRDLL;CGLM_EXPORTS;%(PreprocessorDefinitions)</PreprocessorDefinitions>
       <DebugInformationFormat>None</DebugInformationFormat>
       <PrecompiledHeaderFile />
       <CompileAs>CompileAsC</CompileAs>