- All variables are initialized on declaration, some arrays were not properly initialized
- Static array buffers require memset() for re-initialization on every function call
`RLGL.State.vertexCounter` is a generic counter and it's reused for all `rlRenderBatch`, actually, once render batch is filled, required vertex count is provided through the draw calls, so, the total accumulated count of vertices is not directly registered inside the rlRenderBatch.
`RLGL.State.vertexCounter` keeps that count but one possible improvement(?) could be moving the `vertexCounter` inside `rlRenderBatch` to always keep a register of the total accumulated vertices in that batch (despite that info is provided by the accumulated `draws[i].vertexCount`.
Simplifying, `RLGL.State.vertexCounter = SUM(draws[i].vertexCount)`
The decision to move the counter out of `rlVertexBuffer` is to keep only the data that I think should belong to `rlVertexBuffer` and make it more generic, aligned with raylib `Mesh` structure.
The decision to not add it to `rlRenderBatch` is because it could contain multiple `rlVertexBuffer` and it would be confusing (because it would only register the count of the last filled one).
Instead of registering vertex texcoords and colors on every call, we keep the last defined value and we record everything on `glVertex*()`. Actually that behavior is aligned with OpenGL 1.1 standard.
Note that `extern "C"` calling convention only affects objects that need to be seen by the linker, in our case only functions... but it would also be required by global variables exposed, if any.
Now rlgl uses the `Matrix` type, just make sure it has been previously defined somewhere... I don't like this approach but it's probably the easier one for the users... still looking for a better solution... maybe using something like
`#define MATRIX_TYPE`, so it can be checked in other modules.
rlgl has been redesigned to avoid any dependency to `raylib` or `raymath`, all functions using some of those libs have been reviewed.
- REMOVED: `Texture2D`, `Shader` structs dependency
- REMOVED: `Vector3`, `Matrix` structs dependency
- REMOVED: raymath functions dependency, all required math is implemented in rlgl
- ADDED: `rlMatrix` custom rlgl type
- ADDED: `utils.c`: `rlMatrixFromMatrix()` and `rlMatrixToMatrix()` for a safe conversion between raylib<->rlgl matrix types
- ADDED: `rl` prefix to all `rlgl` structs
- Other small tweaks here and there
- Many functions renamed to follow rl*() convention
- Some internal functions exposed in the API
- Some functionality moved to other modules
- Reorganized all functions by categories
- Make sure it keeps working with OpenGL 1.1 and 2.1
Vr simulator has been moved to core module and completely redesigned. Now user is in charge of stereo-render fbo and also in full control of distortion shader. Code is a bit more complex but better aligned with other raylib examples.
Some enums values have been renamed to be more consistent and also provide a more detailed description:
- ShaderLocationIndex: LOC_VERTEX_POSITION -> SHADER_SHADER_LOC_VERTEX_POSITION
- ShaderUniformDataType: UNIFORM_VEC2 -> SHADER_UNIFORM_VEC2
- MaterialMapType: MAP_ALBEDO -> MATERIAL_MAP_ALBEDO
- PixelFormat: UNCOMPRESSED_GRAYSCALE -> PIXELFORMAT_UNCOMPRESSED_GRAYSCALE
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>
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.
* 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
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.
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.
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.
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
- Mesh, Shader, Material structs have been reviewed to minimize size when passed by value, all required code has been reviewed.
- GetCollisionRayModel() reviewed to avoid pointer, not required because model is not modified inside the function
- UnloadMesh() reviewed, pointer not required
- CheckCollisionRay*() parameters name reviewed
* Working build
* Fix build again, stop deleting files
* Hotfix crash, needs investigating
* Remove VS2015.UWP, I cannot update the project
* Lots of UWP work, added keyboard and mouse press support. Still need to finish scroll wheel, mouse position and cursor hiding, plus other stuff that I haven't seen yet.
* Implemented a ton more things, added BaseApp.h to provide common code to UWP apps.
* Remove constant window dimensions
* Enable and Disable cursor support.
* Actually use mouse delta
* Gamepad Support
* Cleaning and small tweaks
* Restore original example.
* Update comment
* Use 'Messages' to handle the cursor functions so code is more portable.
* Comment
* Comment unused message fields and use vector for mouse pos instead.
* Move messages to utils.h and use messages for everything. No more plat-specific code in raylib.h
* Working build
* Fix build again, stop deleting files
* Hotfix crash, needs investigating
* Remove VS2015.UWP, I cannot update the project
* Lots of UWP work, added keyboard and mouse press support. Still need to finish scroll wheel, mouse position and cursor hiding, plus other stuff that I haven't seen yet.
* Implemented a ton more things, added BaseApp.h to provide common code to UWP apps.
* Remove constant window dimensions
* Enable and Disable cursor support.
* Actually use mouse delta
* Gamepad Support
* Cleaning and small tweaks
* Restore original example.
* Update comment
* Use 'Messages' to handle the cursor functions so code is more portable.
* Comment
* Comment unused message fields and use vector for mouse pos instead.
* Move messages to utils.h and use messages for everything. No more plat-specific code in raylib.h
* Tested some desktop stuff and added projection matrix updates for window resizing.
* Fixed big bad mouse bug
* Fix alt buttons and add hack to combat flickery key presses (far from perfect)
* Remove debug code
* Final commit
* Well, so I thought
* Wow, i am bad
* Remove packages folder
* Remove useless include
* Apply requested changes and fix linux build
* Try to stop packages folder
* Have we fixed the formatting properly?
* Third time's the charm?
* Where did this come from?
* Re-fix
* Autoformat is gonna kill
* Fixed XBOX ONE Support
* Fix tabs
Found some issues when building for web using latest emscripten 1.38.30, traced the error and found that eglGetProcAdress does not return function pointers for VAO functionality, supported by extension.
It requires more investigation but now it works (avoiding VAO usage)
Users can define their custom memory management macros.
NOTE: Most external libraries support custom macros in the same way, raylib should redefine those macros to raylib ones, to unify custom memory loading. That redefinition is only implemented as example for stb_image.h in [textures] module.
Issue was related to vertex tangent attibutes not uploaded to GPU, a quick solution was implemented for new vertex attributes loading for already existing meshes... I don't like it specially but it will work for now.
If we have no data to update/draw, we avoid update/draw. On `DrawBuffersDefault()` if no vertes data is available nothing is drawn but some globals: vertexData, projection, modelview, draws... are reseted.
There shouldn't be any problem if we don't touch those globals in case no vertex have been processed but, just in case, I warn about it.
When drawing LINES or TRIANGLES, vertex are accumulated in same buffer as QUADS and new draw calls are registered but QUADS drawing uses an index buffer for optimization, so, when adding LINES/TRIANGLES vertices we need to make sure next draw calls for QUADS keep aligned with indices buffer.
To get that we just add some alignment vertex at the end of the LINES/TRIANGLES draw calls, to make them multiple of 4 vertex.
When capturing framebuffer, if some element with transparency has been drawn, it retrieves that data... it could be a bit annoying in some situations because we could expect color compositing with background color... It depends on the situation and our needs... but most of the time we don't want that transparency on screenshots.
Specific textures generation code is quite redundant and not flexible for the user, I'm trying to figure out some easy way to allow raylib users to do the same without needing those functions (very specific and shader dependant). RenderTexture loading and Cubemap textures support must be improved.
Problem aligning provided vertex data to multiples of four, because main buffer is intended to bu used with indexed quads... but also shared with triangles and lines.
