* Add support for u8 bone indicies when loading glTF
* Fix segfault for glTF animations not keyframed at 0
When loading glTF animations we lerp between keyframes, and previously
assume that if the frame we are considering has a later keyframe, there
must be a previous keyframe. This is not true if the animation's first
keyframe is some time into the animation. In this case we now
effectively clamp to that first keyframe for any time prior to it.
* Respect parent bones tranform when loading glTF animations
We previously assumed that when loading glTF animations, the bones were
ordered with those higher up the skeleton tree (i.e. closer to the root)
came first in the list of nodes. This may not be true, so now we
repeatedly loop, preparing each level of the skeleton tree one after the
other, starting at the root level. This ensures that any parent
transforms are applied before transforming any child bones.
We also ensure that we have forced the loading of animation data before
attempting to interpolate to generate the animation frames for use
later, without this no animations are applied.
Finally we remove the check that assumed the first node in the nodes
list is the root, and use an invalid index value as the sentinal value
for when a node has no parent. Previously this was 0, which made
distinguishing between root nodes and children of the first node
impossible.
Actually, it seems more logical that rlgl takes care of OpenGL data than the models module...
Also, models module loaded vertex data is unloaded by models module.
* Fixing typecast warnings generated by visual studio 2019 in examples.
* Changes to fixes based on feedback
Co-authored-by: Jeffery Myers <JefMyers@blizzard.com>
Mostly typecasts, and some int to unsigned short changes, and including the windsock 2 library in the examples.
Co-authored-by: Jeffery Myers <JefMyers@blizzard.com>
* Add MUSIC_AUDIO_NONE to MusicContextType and format fixes
- Useful to check the context type to see if the format is recognized. Defaulting to wav causes issues where formats are assumed to
be wav.
* Fix memory issues with LoadMusicStream and UnloadMusicStream
- Set ctxType and ctxData even if the format fails to load.
- Set ctxData to NULL if it fails and check for null inside UnloadMusicStream.
- Change RL_MALLOC when loading formats to RL_CALLOC to prevent undefined behavior.
- Add NULL check when unloading xm file.
* Disabling WindowSizeCallback around fullscreen toggle.
#1294 fixed the issue that toggle fullscreen changes the screen size because of the WindowSizeCallback. The proposed edit by @raysan5 was to comment out WindowSizeCallback which I essentially simplified to disable the callback around the set window monitor functions.
The developers using the ToggleFullscreen function should be aware that they need to size the screen correctly on that same frame. Otherwise they should check if fullscreen and size properly using the GetMonitorWidth and GetMonitorHeight functions.
* Update core.c
Fix issue from merge
This solves issue #1322 which in my opinion was prematurely closed. As the monitor resolution is expected to be the maximum that the monitor supports. My change is that as per GLFW documentation you can get all current video modes sorted by max resolution. When you toggle fullscreen the first video mode returned would be the current screen resolution setup but that doesn't help if you want to know the maximum supported.
* Added example for gltf animation and split some functions for loading model animations into IQM and GLTF similar to how models are being loaded.
* Removed wrongly duplicated function
* Bone loading for gltf model (not working at this point)
* Loading info about vertex to joint connection but animation is still not working
* Skeleton and pose is correctly loaded. Need to communicate about interpolation in GLTF
* The model almost looks like a real person on animation.
* Fixed model loading with bones.
Also updated license info on the model.
* Cleaned up some code and updated examples.
* Fix identation issues
* Fix identation issues
* Fix identation issues
* Documenting the compiler flags
* Moved some android compiler flags and added documentation on them too.
* Some more restructuring.
Removed unnecessary comments that were self described by the code.
Added some more explanations around certain parts of CMake and especially around compiler flags.
CMake relied on this file for configurations and also was interfering in the regular config.h by having a separate definition if building with CMake. This was not entirely correct so instead we will define compile time definitions separately through CMake (CompileDefinitions.cmake) and also will use the provided EXTERNAL_CONFIG_FLAGS that I found that will not use config.h in through the build process.
I also introduced a new compiler option (CUSTOMIZE_BUILD) that when OFF will use the default config.h and when ON will show other options for redefining your own options.
Fixed an error in rlgl.h where if you have both RLGL_STANDALONE and SUPPORT_VR_SIMULATOR you get a compile time error.
This test file is just testing compilation with the library works correctly but is no longer needed because:
- it is not cross platform
- it taps into the CTest system which is better suited for real unit/integration tests
- it can be incorporated into the pipeline of github actions instead in the future
As described in the official documentation https://cmake.org/cmake/help/v3.0/variable/BUILD_SHARED_LIBS.html this flag is global by default and controls if the library will be built as a shared or a static library allowing us to define only one call to the add_library function (without specifying its type). It is also added as an option to be visible in CMake GUI applications.
* Better ignore support for idea projects.
Added a wildcard at the end because different configurations would have a diffeerent build directory.
* Removed external from being a relative include directory for target raylib.
Fixes#1533
* Delete emscripten.cmake
This file is not needed at this point. EMSDK provides a toolchain file that has a lot more things in it and is better supported. Project currently works fine with the documentation provided in Emscripten SDK on how to build projects.
* First pass file separation.
The main two files are cleaner now. Only important things can be seen. Major changes include:
- raylib_static is now the alias instead of raylib
- Repeating segments are removed and pulled into separate files into <root>/cmake
- File is reordered to make more sense
- Installs are better structured
- Library is build into an output directory "raylib" instead of "src"
- All public header files are now set as a public header file
- Source files need to be listed (it is a bad practice to capture them using wildcards and file globs)
- CMakeLists are better commented
* Second pass on the example dirs.
They are quite complex so I'm more hesitant to do major changes. Also it works pretty well. Noticed that I forgot one of the seperated files and added it into src/CMakeLists.txt.
* Returned the header copy as it was convenient to have the public headers copied.
* A better description to the variable RAYLIB_IS_MAIN
Co-authored-by: Rob Loach <robloach@gmail.com>
* Remove debug message
Co-authored-by: Rob Loach <robloach@gmail.com>
* Improvements based on review.
* Simplify the install condition to not be platform specific as it was before.
Co-authored-by: Alexander Neumann <30894796+Neumann-A@users.noreply.github.com>
* Remove some CMAKE variables as they don't affect the build in any way
Co-authored-by: Alexander Neumann <30894796+Neumann-A@users.noreply.github.com>
Co-authored-by: Rob Loach <robloach@gmail.com>
Co-authored-by: Alexander Neumann <30894796+Neumann-A@users.noreply.github.com>
* Don't create an ortho matrix when the viewport is 0 in any axis.
Not all compilers divide by 0 and return inf, some segfault.
The matrix is not used by anything when minimized, so it just needs to not be called.
* Better fix that always ensures the rlgl matrix is always valid
* Better fix that always ensures the rlgl matrix is always valid
* Added a new utility function to get the current monitor.
Some of the other functions rely on passing on a monitor id but we only have a function for querying the count of monitors available.
* Updated the comment to correctly reflect the function action
This monitor should even probably be passed as an id instead to make it more consistent with the rest of the API but this would break existing applications so I am hesitant to change the function signature directly.
* Add a function to know if two lines intersect each other and if yes, get the intersection point
* Remove indents
* Rework the declaration of the 'Vector2LineIntersect' function, and add the 'Vector2SegmentIntersect' function
* Remove bad indents
* Fix compilation issues
* Fix compilation error
* Fix compilation error
* Replace keyword '_Bool' by 'bool'
Pefixing/Suffixing some data type identifier in the variables is not a convention used in raylib, just reviewed it for consistency...
Still, I kept the *Rec suffix in some functions.
Previous implementation did not consider any padding while drawing the characters on screen (despite being available on the font atlas), so, only minimum character area was drawn.
If some text effect shader was required (shadow, glow, outline...), there was no space in the drawn quad to draw that pixels effect.
This commit corrects that issue.
This is a BREAKING CHANGE!
To address the linked issue, several functions have been renamed and couterpart functions have been created to free loaded memory:
- RENAMED: GetImageData() -> LoadImageColors()
- RENAMED: GetImagePalette() -> LoadImagePalette()
- RENAMED: GetWaveData() -> LoadWaveSamples()
- ADDED: UnloadImageColors()
- ADDED: UnloadImagePalette()
- ADDED: UnloadWaveSamples()
Previous GetKeyPressed() method was actually returning unicode codepoints equivalent values instead of the key-code of the pressed key. So, it has been replaced by GetCharPressed(), returning unicode codepoints and GetKeyPressed() now returns key-codes.
* Changed extension for web from .bc to .a
I did this to support vcpkg expectation. When using the library from vcpkg for web you would install it using vcpkg install raylib:wasm32-emscripten but also vcpkg expects the output lib to be with .a extension instead of .bc
Doesn't make a difference for standalone builds or when raylib is used as a subdirectory dependency.
* Added headers to install targets to fix vcpkg usage
Currently vcpkg works by installing the cmake target into its packages directory. The problem is that install only copies the public header at this point so we need to add the others to the install targets.
I did this to support vcpkg expectation. When using the library from vcpkg for web you would install it using vcpkg install raylib:wasm32-emscripten but also vcpkg expects the output lib to be with .a extension instead of .bc
Doesn't make a difference for standalone builds or when raylib is used as a subdirectory dependency.
* Fixed the build for web using CMake.
I found that the build for me was failing and I added some if defined checks in the core.c file where the glfwSetWindowAttrib was used. (error: implicit declaration of function 'glfwSetWindowAttrib' is invalid in C99 [-Werror,-Wimplicit-function-declaration])
I also changed some values in the toolchain file so that it correctly uses the .bat files when on windows.
* Cleaned up the additional variables (they are not important)
* Added more improvements to cmakelists
Added the option to use the system provided Emscripten toolchain to be more uniform with other libraries.
Fixed and issue which prevented example being built from cmake and also building with html extensions properly.
* Fixed ENUM to STRING because of a missed warning
WARNING: Several functions removed, replaced by SetWindowState() / ClearWindowState() equivalents, only for advance users.
ADDED: ClearWindowState() to reset window state
REMOVED: HideWindow() / UnhideWindow()
REMOVED: DecorateWindow() / UndecorateWindow()
There were some problems about frameCount vs sampleCount that could cause some breaks.
raylib audio structs stores sampleCount = frameCount*channels.
Most libraries return framesCount instead of sampleCount.
stb_vorbis seems to refer to framesCount as samples.
All required functions have been reviewed.
There is no zero-check, so window upscales to nothing.
SetupFramebuffer() is kinda wrong, it uses not its params, but global variables. I won't touch it, maybe it has purpose
Vector3Reflect exists but not Vector2Reflect. The code is pretty much the same.
I'm not sure what RMDEF does, but I added it to match other function definitions (haven't done much C programming, maybe I'm missing something). Can someone explain to me what it does?
* updated README.md
* fixed CMakeLists.txt to allow building and debugging with Visual Studio Code and CMAKE Tools extension
* added PLATFORM_DRM
contains mouse pointer code from https://github.com/chriscamacho
* removed redundant cleanup in InitGraphicsDevice
* fixed DRM connector mode selection
* added choosen DRM connected mode to log output
* added respecting TargetFPS on DRM mode selection, default to 60
* added support for GetMonitorRefreshRate
* changed SUPPORT_MOUSE_CURSOR_RPI to SUPPORT_MOUSE_CURSOR_NATIVE
* changed avoidProgressive to allowInterlaced
* cleanup, function extraction and improved mode selection
* README reverted to original for PR
* line endings fixed for core.c
* removed old code
* mouse pointer reverted to small square
* replaced SetGraphicDeviceName() by DEFAULT_GRAPHIC_DEVICE_DRM
Co-authored-by: kernelkinetic <kernelkinetic@outlook.com>
Function has been reviewed to avoid any direct OpenGL call and use rlgl functionality, also, GenDrawCube() has been replaced by the internal batch system with DrawCube().
WARNING: rlEnableTexture() call must be issued after enabling the current framebuffer when using batch mechanism because it includes a set of security checks to avoid batch overflow and push/pop matrix operations.
* fixed wrong error message for input device
* error message without errnofor input device on RPI
to prevent from including additional header
Co-authored-by: kkl <klingenberger@bfmc.de>
This redesign allows more flexibility when creating RenderTexture and a simplification (and hopefully removal) of `GenTexture*()` functions, that should not belong to this model but the user code, due to the use of custom shaders.
Also, this new API opens the door for a possible GBuffers type and advance rendering possibilities...
Some functions of the API have been also simplified or even removed.
rlgl module can be used as an standalone library, so, a version for the library has been added: v3.1.0, matching current raylib version.
Before deleting certain objects, they must be detached from their parents. That's the case for shader objects after linkage to shader program and also for the texture/cubemaps/renderbuffers attached to framebuffers. If objects are deleted before detached, they are kept in memory to avoid accessing deleted data.
rlBlendMode() has been added to rlgl to be used on BLEND_CUSTOM. This functionality is exposed to advance users.
In any case, new blending modes could be added if required.
When using USE_PTHREADS=1 on libraylib.bc compilation, program requires --shared-memory passed to the linker and despite it compiles, it fails on execution (at least for me).
There are no *.lib files that could overwrite each other when building
both static and shared versions of raylib at once. So just use the
normal library name without _static suffix.
When you install emsdk out of the box, the directory is no longer `$(EMSDK_PATH)/python/33.7.4_64bit` but rather `$(EMSDK_PATH)/python/3.7.4-pywin32_64bit` noting the `3.7.4-pywin32_64bit` change on the last file.
Thus, without adjusting this in the raylib Makefile it will throw an error:
```
PS C:\raylib\raylib\src> make PLATFORM=PLATFORM_WEB -B emcc -c core.c -Wall -D_DEFAULT_SOURCE -Wno-missing-braces -Werror=pointer-arith -fno-strict-aliasing -std=gnu99 -Os -s USE_GLFW=3 -I. -Iexternal/glfw/include -Iexternal/glfw/deps/mingw -DPLATFORM_WEB -DGRAPHICS_API_OPENGL_ES2 '"python"' is not recognized as an internal or external command, operable program or batch file.
```
However, changing this line in the Makefile compiles raylib for web as expected
`"raylib library generated (libraylib.bc)!"`
Reason for removal: This function forces a specific Image data format, it copies data internally (it could be confusing and lead to memory leaks), it's redundant, there is a simpler alternative and raylib promotes using structures directly
It seems some internals change recently due to web security reasons and some emscripten HTML5 funtionalities like pointerLock or fullscreen modes behave very weird or just don't work as expected
Despite all the effort put on function optimization, dealing with alpha blending is complex, considering src and dst could have different pixel format...
Several functions used to load external files have been reviewed to just load the full file with LoadFileData() and load data from memory from there. This way all file access functionality is centralized in utils module.
Functions reviewed: LoadDDS(), LoadPKM(), LoadKTX(), LoadPVR(), LoadASTC(), SaveKTX()
* Adding Norm and Remap functions
// Normalize input value within input range
// Remap input value within input range to output range
* Rename Norm to Normalize
To make it uniforms with Raylib's functions
* Calculate Remap without other functions
Current looping system was broken, `loopCount` has been converted to `bool looping` and user can enable/disable with `music.looping = false`. `SetMusicLoopCount()` has been removed.
Most warning were related to types conversion (casting required) and unsigned/signed types comparisons.
Added preprocessor directives (_CRT_SECURE_NO_DEPRECATE; _CRT_NONSTDC_NO_DEPRECATE) to avoid warnings about unsafe functions, those functions are safe while used properly and recommended alternatives are MS only.
Some external libraries still generate warnings.
Created RenderBatch type and related functions to allow rlgl users creating custom and separate render batches and not only rely on one internal render batch.
This feature has not been exposed on raylib yet because it implies some breaking changes. Right now a custom RenderBatch is just used internally in rlgl.
There are multiple #define values around raylib, usually not exposed for redefinition, just reviewed all of them to allow users redefining them on compile time if required.
Also, multiple #define have been renamed and commented.
* First draft of UWP rework.
* Read desc
- Moved UWP specific functions to uwp_events.h
- Removed BaseApp.
- Implemented example UWP lifecycle.
* Added GIF recording and screenshot support.
* Character inputs and filesystem stuff
* Fix game closing on Xbox when B is pressed.
* Fix the gamepad binding hack
* Add as many keys as I believe are possible.
* Implemented mouse locking of a sort.
* Remove rogue todo, the rest are for a game dev using this example.
* Implemented touch how I "think" it should work. I cant test this.
* Review.
Optimizations of allocators.
Renamed 'Stack' to 'Arena'.
Replaced certain define constants with an anonymous enum.
Refactored MemPool to no longer require active or deferred defragging.
Actually this function is useless and potentially dangerous, internally, a shallow copy of data was done. Same could be accomplished accesing image.data directly.
Now it's blazing fast. It modifies image.data directly.
Previous version could take up to 3-4 seconds to draw a line... now it takes some nanoseconds...
As noted in #539, the symlink check was to support compiling raylib on
VirtualBox vboxfs as no symlinks are supported there.
With me using Linux as daily driver outside VirtualBox now, this 'feature'
is most likely unused by anyone, but still annoys other users (#791).
Just revert it and be done with it. If there are really others building
raylib on vboxfs or a similarly crippled file system under Linux, they
should now step forward when their build breaks.
This reverts commit b8ca51fd01.
- Added NULL return if the replacement string (`by`) is empty
- Reordered sanity checks since there's no need to initialize variables if the strings are invalid.
Errores fixes:
models.c,2843 Either the condition 'fileData!=NULL' is redundant or there is possible null pointer dereference: fileData.
raudio.c,805, 806, 807, 808, %d in format string (no. 2) requires 'int' but the argument type is 'unsigned int'.
* [cppcheck] Improvements in SaveStorageValue() in core.c
in file core.c cppcheck shows errors only in function SaveStorageValue():
* Common realloc mistake: 'fileData' nulled but not freed upon failure
* Memory pointed to by 'fileData' is freed twice.
Validation:
* Tested examples/core/core_storage_values.c
* Launched Unit Test for this function
* Rerun CPPCHECK afer fix
* [cppcheck] Change functions header to accept only positive position in files
Changes:
* Functions SaveStorageValue(), LoadStorageValue() (core.c)
* Functions LoadFileData(), SaveFileData() (utils.c)
* Headers in raylib.h
Validation:
* Tested examples/core/core_storage_values.c
* Launched Unit Test for these functions
* Rerun CPPCHECK afer fix
* Update compilation warning flags
Removed the -Wall flag that allow many warnings and added manually all the warning flags that don't generate warnings in the src/external folder.
Specifically, these are some of the flags reporting errors in src/external:
-Wmissing-prototypes
-Wunused-variable
-Wunused-value
-Wunused-parameter
-Wunused-function
* Uptade compilation flags in examples with missing -std=c99
* Android: Better track touch input returned from IsMouse*()
Switch to actually tracking touch input to use for "mouse" input rather
than the gestures system. The gesture system as an abstraction ontop of
raw touch input loses some information needed to map to "mouse"
input.
Before,
- IsMouseButtonReleased() triggers immediately after the initial touch
(because GESTURE_TAP activates immediately on touch) instead of waiting for the
touch to be released.
- IsMouseButtonUp() returns false, when it should just be the opposite
of IsMouseButtonDown().
- IsMouseButtonDown() returns true only after GESTURE_HOLD (which
activates after some period of time after GESTURE_TAP), when instead it
should just be true whenever there is touch input i.e. gesture !=
GESTURE_NONE or alternatively when any input is received on the screen.
After this PR, touches map closer to mouse input.
- IsMouseButtonReleased() triggers when touch is released (last frame
was touched, this frame not touched).
- IsMouseButtonUp() returns the opposite of IsMouseButtonDown()
- IsMouseButtonDown() is true when
(AMOTION_EVENT_ACTION_DOWN || AMOTION_EVENT_ACTION_MOVE) and false when
(AMOTION_EVENT_ACTION_UP)
* RPI: Include index check for RPI in GetTouchPosition()
We might want to customize this in feature for multi-config builds (e.g.
a Visual studio build with both Debug and Release configurations).
Output the variable value for user awareness.
This was working in 2.6 but no longer does in current git tree.
It appears touch position is only tracked on
AMOTION_EVENT_ACTION_[DOWN|UP], which only registers the initial touch
on the screen. Subsequent movement is not tracked into CORE.
Touch position and the Gesture System appears to be updated twice in
AndroidInputCallback in what looks like perhaps a copy paste error (code
is identical) with the exception of tracking AMOTION_EVENT_ACTION_UP in
the 2nd copy of the code (but this is not necessary to track).
If you need to track the first touch or release touch position, you can
do so with checking IsMouseButton[Pressed|Released] on the same frame.
This patch makes it so the touch position is always updated, and merges the
duplicated code into 1 singular code path.
NOTE: This functionality has an important issue, line space is fixed to 1.5 font.baseSize pixels, depending on the font and how it has been generated that spacing could be too wide or too narrow...
In `InitGraphicsDevice(...)`, the Android section has a screen security check
like other platforms- but CORE.display.width, CORE.display.height are
not set yet, so the security check sets it to 0. So ensure we query the
device's screen width and height before the screen size security check.
This also gives you the ability to run a proper full-screen application
on Android without any scaling and guess work on the target device by
setting screen width and height to 0 in `InitWindow(...)` and using
`GetScreen[Width|Height]()` to get the actual values.
* Update GLTF support to include loading color for albdeo (saved in the color value of the materialmap), support occlussion and emmission maps.. as well as some quality of life updates.
* clean up to use single image
There could be some cases where we need to define the internal audio buffer size, now it's possible.
RENAMED: InitAudioBuffer() to LoadAudioBuffer()
RENAMED: CloseAudioBuffer() to UnloadAudioBuffer()
Renamed some defines and reviewed some comments.
Just for clarification, no plans to remove libc dependency, just did some code analysis to see how much raylib depend on stardard C library. My conclusions:
- stdlib.h: primary dependency is for malloc() and free()
- stdio.h: primary dependency is for FILE access, maybe it could go through a custom ABI?
- string.h: just around 8 functions required
- math.h: just around 8 functions required
- others: 1-2 functions required for some other headers
* Update to miniaudio 0.10
This replaces the old ma_pcm_converter with ma_data_converter.
At this time of this commit, miniaudio 0.10 is still in the testing
phase. To make it easier to update miniaudio.h during this period, I've
temporarily moved the @raysan5 Win32 customizations to raudio.c because
there may be quite a few updates to miniaudio.h during this time.
* Use miniaudio's built-in volume control.
Short non-looping sounds can sometimes think they need to keep playing and will output their first few frames again. This helps to break out of all the loops instead of just this one.
* We get the video mode from the target monitor and use that to set, therefore
keeping windowed-fullscreen
* Added a GLFW_AUTO_ICONIFY 0 hint so that glfw does not minimize the window
when changing focus from a windowed fullscreen window. This is more expected
behavior for windowed full screen, when a user alt-tabs or clicks on a window in
the other monitor, they expect their windowed-fullscreen window to remaining up.
We were doing this before, but it was deleted during the
last GLFW update. Readd it to fix the associated macOS CI failure.
Fixes: ea5cd42e6 ("Update GLFW to version 3.4")
If press/release events for a mouse button come too fast, then using
`IsMouseButtonReleased()` does not work. This has been noticed when
using a touchpad on Linux when tapping with two fingers two emulate
right mouse button click.
The situation looks like this:
```
BeginDrawing <-- current==released, previous==released
Pressed <-- current=pressed
Released <-- current=released
IsMouseButtonReleased <-- returns false because current==previous
EndDrawing <-- previous=released
```
The fix is to update the previous mouse button state in addition to
current mouse button state when `MouseButtonCallback()` is called by
glfw. Now the situation is as follows:
```
BeginDrawing <-- current==released, previous==released
Pressed <-- current=pressed, previous=released
Released <-- current=released, previous=pressed
IsMouseButtonReleased <-- returns true because current!=previous
EndDrawing <-- previous=released
```
