raylib 2021 Wishlist!

[raysan5]

Just opening a small discussion to get feedback from the community about new features/improvements/ideas for raylib in 2021.

Please, try to be proportionate, keeping in mind the raylib core values.

Also note that shared ideas in this post are not meant to be implemented (at least not all of them) but they would be a great reference to know what raylib users would like to see in the library for the next year.

[RobLoach]

2020 was a incredible year for raylib. So much has changed since 2.5.0 last year.... Nicely done, @raysan5! 😉 ... Here's my wishlist, in order or priority.

Additional *FromMemory() functions

There is support for loading Images, Fonts, and Waves from memory, but I believe there is a good opportunity to extend that to Textures, Models, Meshes, Materials, ModelAnimations, Music, Shaders, etc. That would allow raylib to be run without needing direct file system access. rres is a thing, but I think it's valid to have the support directly in raylib itself.

• [Models,Materials] Create LoadFromMemory versions of all Loaders)

Expanded Software Rendering

The software rendering allow support for a whole bunch of devices that don't have a graphics processor. Would love to see that expanded upon across some examples.

• [core][texture][text][shapes] Software renderer for 2D

Semantic Versioning and Frequent Releases

Fully adopting semantic versioning fully will allow more frequent release cycles, and achieve better stability across applications using raylib. Major.Minor.Patch Any breaking changes queue a major version. Any new features that don't break backwards compatibility queue a minor release. Any bugfixes queue a patch version.

[raysan5]

Very nice, detailed and proportional wishes! Thanks! Just one comment about versioning, what about Major.Minor.Commit? Completely agree about more frequent releases, I'm working on it... 😉

[orcmid]

(I just saw this and don't know if there was a response.)

I recommend Semantic Versioning (version 2), aka semver2.

There's additional information on identifying builds, platforms, etc., that fit into release packaging. I'm uncertain about Commit numbers/codes.

[ctrl-shift-make]

u could add opengl embed in gui

[raysan5]

I don't understand what you mean...

[ctrl-shift-make]

I mean like qt. You could get an openGL scene/view in a gui

[ctrl-shift-make]

In raygui module of raylib

[raysan5]

raygui works in a completely different way than Qt... but you can already do that, just use a RenderTexture to render your OpenGL scene and then draw the texture on the screen...

[ctrl-shift-make]

kthx

[pmprog]

Support for multiple windows has been on my mind.
Allegro5 supported this, and it lets you take advantage of multiple monitors on a single computer.

Can windows be resized too? I can't remember now

[tokorv]

Windows can be resized see this example.

[UC101]

A multi-window mode is certainly something I'd be interested in.

[JonSnowbd]

Thanks for all the cool new features!

As for requests I'd love the ability to supply RayGUI with an artifical mouse position replacement for worldspace gui and to correct mouse input when using a black-bar style rendertarget to render your game scaled.

[raysan5]

Mmmh... you can do that providing your own GetMousePosition() definition when RAYGUI_STANDALONE.

[JonSnowbd]

Thats a good idea, I'll try that.

[zzador]

1. Loading of meshes (at least obj files)
2. General creation methods for "Mesh" like CreateMesh(numVertices, numNormals, numTexCoords, numIndices, ...)
3. Additional CheatSheets for raymath & rlgl

[raysan5]

@zzador I'm afraid OpenGL ES 2.0 does not support 32bit indices, to remain multiplatform with same code I had to limit that option.

[zzador]

@zzador I'm afraid OpenGL ES 2.0 does not support 32bit indices, to remain multiplatform with same code I had to limit that option.

Oh...I didn't know about that. Well, if I would implement it myself I would clearly break the multi-platform support that is given by raylib (write once, compile everywhere) or I would implement it on every platform. That means that most android systems out there only support indexed meshes with up to 65536 vertices...that is sad. :( I need to find a workaround...maybe splitting my meshes.

[JeffM2501]

Break your mesh up into multiple objects.

[zzador]

Okay, that worked.
Here's another wish: MultiThreading support!
I'm currently using PThreads for MT (even on windows). Would be nice to have that support directly in raylib. In the simplest form that would only be one additional function like:

Thread StartThread(ThreadFunc func);

[chrisdill]

I opened a issue related to point 2. Feedback is appreciated.

[assyrianic]

iOS support with a Metal API implementation for rlgl

[raysan5]

I'm afraid that's very unlikely...

[Mea-sama]

In Raylib_cs it would be nice if the Raylib objects would be Serializeable.
Saving a Color, Rectange or else struct in other objects greates the opportunity of individual objects that can be saved without having to convert the data to windows objects.
SerializableAttribute Klasse

I know the main focus lays on C, not C#, still i think that is a pretty easy to implement feature to help C# programmers saving gameprogress easier

[raysan5]

I think this request belongs to raylib-cs project.

[DunkUK]

Exposing more of the MiniAudio API to improve the audio feature set - for example, adding a SoundInstance object so we can change the volume and pitch of individual sound instances, as well as set the left-right channel mix (i.e. panning/positional audio). Positional audio could be tied to Raylib's camera, but users might enjoy controlling this part themselves.

Consider re-testing/debugging audio for emscripten, as PlaySoundMulti has been known to behave unreliably in different browsers.

[raysan5]

This is on the list.

[a740g]

@raysan5
Couple of observations / feedback / suggestions for the audio subsystem:

1. I am quite certain that the audio mixer does not run on its own thread by default. Do you think it would help to run the mixer on its own thread? Again, this is from my quick study of the library, and I may be wrong
2. The MOD and XM streaming decoders do not use interpolation (trusting my ears) which make them sound low quality
3. XM playback is not perfect, and some songs simply sound awful. This one for example sounds bad (in ways I cannot really explain) https://www.modules.pl/index.php?id=module&mod=5835. I compared this with https://github.com/8bitbubsy/ft2-clone and could immediately hear the difference
4. S3M & IT are also popular tracker formats. Adding support for these would be huge bonus
5. OPUS support for LoadMusicStream

[ghost]

Vulkan support

[raysan5]

Maybe in a future...

[arydevy]

I will love to see this!

[jackwlee01]

This would also allow Raylib to run on Metal/iOS/Mac via MultenVK.

[advayta]

I am novice and just started exploring and learning so ignore if following is already achieved or achievable and left for programmer to code as I am struggling to find nice documentation or help this could be just me!

1. Native screen resolution detection (currently requires glfw3 or similar library to get the detail)

// GLFW3 to detect video mode and monitor detail
    int count;
    GLFWmonitor* m;
    const GLFWvidmode* modes;
    
    if (!glfwInit()) {
        // exception handling - function assumes the call will be success everytime, returns 0 if error
    }

    m = glfwGetPrimaryMonitor();

    const GLFWvidmode* currentMode = glfwGetVideoMode(m);

    modes = glfwGetVideoModes(m, &count);
    
    // If not terminated pointer memory access violation occurs as raylib internally refers to the same location
    glfwTerminate();

    //GLFW3 ends

2. Texture/Image scale to fit/stretch the resolution e.g., 640x480 screen resolution (4:3) with tile map of 100x100 tiles with 16x16 tile size once rendered and programmed for this resolution, it should automatically scale to adhere to any resolution/ratio e.g. 1920x1080 16:9.

For example, following background will be good for static 800x450 resolution and will look distorted or misaligned for any higher or other resolutions set e.g., 640x480, 1440x900 etc. with aspect ratio 4:3, 16:9, 16:10 or 21:9.

Texture Background Demo

3. Detailed Documentation with examples for new players like me beside the cheat sheet as that would be more like a reference to experts in what they are doing and they know well

[pmprog]

This is what i use to get get the Native screen resolution

	InitWindow(0, 0, "");
	int monitors = GetMonitorCount();
	screenresolution.x = GetMonitorWidth(0);
	screenresolution.y = GetMonitorHeight(0);
	CloseWindow();
	InitWindow(screenresolution.x, screenresolution.y, Title);

Hope that helps

[advayta]

Wow! this was so easy, I wasted hours to get GLFW3 compiled and worked to get to the same path, thank you!!!!

[pmprog]

Yeah, didn't really understand whya window was needed to get monitor information. That threw me at first, but anyway.
From there there's no automated tilemap, but they are easy enough to write.
Just create your 1600x1600 texture, draw your tiles into it and then render just a 640x480 into your display.
Alternatively, you could just render the tiles required to your screen.
The first approach is probably better if your tiles are fairly static, but will consume more memory.
That said, I don't really understand how raylib (or even GPUs) handle batch rendering, which lets you render multiple images ina single draw call. Might be even less to be saved with the buffered texture approach

[raysan5]

InitWindow() does a bunch of initializations, actually, only glfwInit() is require to get that information before Window creation but exposing just that function would require some internal redesign...

[YusufSuleman]

JSON and XML.

[raysan5]

You can use external libraries to support JSON and XML file formats.

[bloodhaze13]

Some Ideas:

IOS build example.
I know that OpenGL support for Mac/IOS is deprecated, but considering the number of OpenGL apps for IOS I believe there is a good chance that there will be some viable compatibility layer (like MoltenGL) which will allow apps to continue to function. I have access to IOS/Mac devices so might even be able to help out with this.

2D Collision Masks
I used these a bit when recently trying out Pygame (https://www.pygame.org/docs/ref/mask.html) before moving across to Raylib. Might even be able to get some inspiration from Pygame source since the implementation is in C: https://github.com/pygame/pygame/blob/d7a76e0fc5cc92cddbcdbce58311773e311bbb6d/src_c/mask.c#L761

Examples of the GetCollisionRay... collision functions
I was not able to find these though they may exist
-RayHitInfo GetCollisionRayMesh
-RayHitInfo GetCollisionRayModel
-RayHitInfo GetCollisionRayTriangle
-RayHitInfo GetCollisionRayGround(Ray

Additional 3D Camera Types

• Camera fixed to grid with only 90 degree turns like some old dungeon crawlers (https://github.com/LutzGrosshennig/unity3d-AnimatedGridMovement-Camera)
• Free flying? Even something simple like https://pastebin.com/WDXFQeh3

edits: removed inline code and added link to pastebin

[raysan5]

@bloodhaze13 just noticed I didn't answer you message.

About iOS support, I don't have any macOS device, they are required to develop and test... but after OpenGL deprecation, it's not a platform I'm interested in.

About collision masks and additional GetCollisionRay...(), I think that functionality can be better implemented in user side instead of inside raylib.

About additional 3d camera types, actually, camera system should be probably removed from raylib and let the user manage it. raylibExtras includes a better first-person camera than the one provided by raylib.

[lileyDatrix]

Suggestion:

• Provide a more accessible method of upscaling windows.

I've just shifted over to raylib from SDL2 and this is one of the first few things I had to test.
In my opinion, I shouldn't have to render a separate texture to the game window just to scale (for example) a 320x240 game to a 640x480 window. While I'm thankful that this all has to be written once per project and that the examples provided are extremely helpful, my primary complaint is that it's convoluted compared to how SDL2 handles window upscaling. Perhaps a function similar to SDL_RenderSetLogicalSize() could be implemented.

[JeffM2501]

That function in SDL2 just sets the viewport size of the backend rendering system it uses for shapes. The texture method usually represented for raylib is more flexible and offers more scaling options, but is not the only way to do it.

You could also emulate the same behavior SDL does in raylib by building an orthographic camera with the same viewport parameters and drawing to the screen like that. You won't get as many filtering/smoothing/alignment options as the texture method, but it would be a simpler process since it would just be a single begin/end call with a single setup. For SDL2 with an OpenGL backend, SDL_RenderSetLogicalSize will literarily just trigger a viewport change to the orthographic projection matrix the same as can be done in raylib with a Camera setting.

I don't think this should be part of the core library as there are too many different ways to do it, each with different benefits and drawbacks and you don't want to force the user into one specific method. If there is need to "hide" the mechanics of render buffers from the user, then it would be better serviced by an optional library.

[jmorel33]

The wishlist is great and it looks like a good portion is already done!

Personally I would like a more recent version of OpenGL supported in order to get the benefit of more advanced shaders and possible performance improvements before Vulkan.

Also support for more Tracker formats :)

I think opening up the ring buffer architecture would be great for designing soft synths and things likek that, or to use outside of the audio realm for example I wish to reroute the console to a memory space and you need a ring buffer structure for this. Maybe that makes sense I never tried it. But I need to grab the different console outputs and use that to live debug my running app.

[zaher]

I like to have mid files support especially from memory I want to generate it on the fly from mmf music language.

[JeffM2501]

Not everyone needs or wants midi. This should be a sperate library in order to not add maintenance burden to the raylib core. I would argue that xm and mod support should also be removed from the core and made an external library to keep the core of raylib small and maintainable.

[raysan5]

@zaher I considered supporting MIDI files, the actual issue with MIDI is the requirement for a sounds bank to play them, so, it would require adding one to the library just for MIDI... or at least adding specific API functions to configure one external... at that point, it's probably easier to just let the user manage MIDI loading and playing.

@JeffM2501 All file-formats supported by raylib can be disabled with the config flags at compile time, also disabling the related code/external libs inclusion. Actually, most of the external libraries used by raylib are just to load different file formats.

[lauchimoon]

Networking module! :D

[raysan5]

rnet.h module tried to be that addition, I designed a possible API and a more experienced user than me implemented it but it was abandoned at some point. I'd like to have a simple network module on raylib for basic stuff, similar to audio module. Here it is the open issue: #753

Lately I saw nbnet library and it can be a good candidate to be wrapped by rnet.h module. In any case, it requires some work because I'm not an expert on networking...

[Shibulski]

Have you looked into "asio c++ library" (https://think-async.com/Asio/)? It is header only with low dependencies.

[raysan5]

@Shibulski I haven't look at it but it's C++, raylib is C

[Shibulski]

Sorry, you are right. I forgot that C thing :)

[ghost]

What about ENet though?

[ghost]

Raylib as a single file header?

[raysan5]

Usually, libraries that fit in the single-file header-only pattern are libraries that only depend on the standard C library (or system libraries), so, they are completely portable and easy to integrate in that format but raylib already depends on multiple other libraries so, I think that approach is not the best for raylib on its current state.

[SasLuca]

I might be able to get this working at some point, its difficult but possible, but since I am very busy I can't make any promises right now.

[overdev]

Since this is a whishlist, don't mind my little whish, please:

3 functions:

• one that calls the OS file open dialog (allowing to select multiple files);
• idem for save dialog (is it the same for opening files?)
• idem for folder selection

I'm not ignoring the utility of GetDroppedFiles(), though.

[zzador]

The OS File dialog is resolution dependent. In other words it will f**k up fullscreen mode if you are currenty in it. Had the problem over the last weekend. The solution was ToggleFullscreen -> FileDialog -> ToggleFullscreen, but it is very ugly. I think you should look for a non-native FileDialog in RayGUI instead.

[raysan5]

@overdev For my tools I use tinyfiledialogs library that works very well on Windows/Linux/macOS, for systems that do not expose modal dialogs easily (HTML5, Android, RPI-native...) I implemented custom ones, you can see the implementation example on my tool rFXGen

[overdev]

@zzador I was trying to guess the issues related to this wish, but I preferred to read from you guys who knows better. You're right, toggling fullscreen mode to show a dialog is not cool. Thanks for the info. 👍

@raysan5 the C/C++ world is full of good surprises. I became the kind of programmer that does everything I can in Python (the way it goes, soon I'll probably be eating, drinking, sleeping in Python too...) so, if there is some .dll/.so/.dylib + .h files available, it's almost always enough. It seems this tinyfiledialogs is exactly what I'm looking for. Thanks! 🚀

[Shibulski]

My Wish: someone already mentioned this:
Support for Metal (since opengl is depcrecated on iOS).

[raysan5]

I'm afraid this is not going to happen. Metal requires Swift language and a bunch of Apple devices to develop/test.

[Shibulski]

I understand, it was just a wish ;) Maybe somewhen in the future. You could use the "c-99"-part of "bgfx" which already implemented metal (and vulkan, etc.).

[tapir]

can we have vulkan backend? that way people can also use moltenvk.

[Gavin-Williams]

Can you please provide a nuget package. Raylib is difficult to setup. In fact, I don't know anyone who has actually set it up from scratch, everybody has had to use a pre-existing template. And even then, it doesn't always work. Why? Nuget packages in C# are so easy to reference and they just work.

I've just copied a template, and it doesn't run, because 'cannot find include file raylib.h'. I searched for a nuget package and there are only packages for C#. I know it's probably straight forward to some, but I come from a world where header files and includes are not a thing, all that garbage is managed by the build tools (as it should be). And it just makes me not want to use it when I have to search the internet for header files and dll's. It's so antiquated. I have no interest in the build process, I want to simply select a package from a list and like magic, there it is, I can use it, and I can get on with the business of working on my project.

I've read mentions of Make ... I'm not gonna use that thanks. I just want a standard VS build.

[raysan5]

@a740g I have no plans to release and maintain a Nuget package, actually I don't know how they work... but if you want to contribute with it and become the official maintainer I will be very pleased, raylib is an open-source project open to everyone to contribute and support their contributions for some time.

EDIT: You mention The Allegro team but I'm afraid I'm mostly alone maintaining raylib.

[a740g]

Got it @raysan5

I'll check if I can find some time to do it.

[a740g]

@raysan5
Here is the first (probably third) version of raylib Nuget package for C. It is now published on the nuget.org https://www.nuget.org/packages/raylib/
Anyone should be able to search and install the raylib Nuget package directly from VS2019 & VS2022.

I had a rollercoaster ride with this one with embarrassingly limited Microsoft documentation for creating Nuget packages for C.
Currently, the package defaults to using the DLL version of the library. There is a custom raylib "Library Type" option that can be access from project properties. One can use this to select the static library.

However, trying to build using the static library throws a ton of link errors.
I am attaching the output log below. If you get some time, please look, and let me know if you have any ideas.

output.txt

Thanks a ton!

Edit: Here if the GitHub repo https://github.com/a740g/raylib-Nuget

[raysan5]

@a740g Oh! That's very nice! Great work!

About the issue with the static library (raylib.lib), it seems related to how it is built, here I found some related info: https://stackoverflow.com/questions/27340218/adding-nodefaultliblibrary-when-building

[a740g]

Fixed! I was missing a %(AdditionalDependencies) in raylib.targets. All I needed was a little coffee. :)
Repackaged and published on nuget.org

Thanks @raysan5

I'll will try my best to keep this updated with your releases.

[jmorel33]

Want more miniaudio integration;

• Use of basic waveforms as provided by miniaudio; square, saw, sine, triangle, noise
• Use of filters; low pass, band pass, high pass, notch
• control of EQ and balance

Essentially allow sound minded people to play around just like graphics minded people :)

[adct-the-experimenter]

I second this. I would like to apply filters to sound to make a dynamic sound environment in a game.

I am willing to help with this.

[jmorel33]

We can work together on this if you wish :)

Also I can share with you a synth engine someone made using RayLib and streaming the generated audio. It's also using ImGUI to tweak parameters.

I am not a big ImGUI user myself.

[adct-the-experimenter]

Yes! Let's work together on this. Please share the link to the synth engine made using Raylib.

[jmorel33]

This is the synth code https://github.com/perky/coding_a_synth_in_c

I found it overly complicated to actually read, so I spent a few hours to study it and simplify it.

The issue right now is it's tied to working using a MIDI keyboard, so it uses Windows to scan for all the midi devices first, and I made it so you can choose any of your MIDI devices by typing the number and enter.

and my forked version

https://github.com/jmorel33/coding_a_synth_in_c/tree/main/synth

Since this was a tutorial synth how-to, I want to pull it away from that and actually make it a full fledged synth you can play around with. Still currently a lot of work needs to be done, but I at least figured out how the velocity is implemented, and it can do some pretty nice modulation tricks.

It is not designed in a very conventional way which is why it requires lots of work to fit in a usable way for a game engine.

[adct-the-experimenter]

Thank you for sharing the code. I am going to fork your version and figure out how it works.
Also try to figure out applying filters with raylib miniaudio implementation.

[Jeff425]

Suggestion to add a way to draw models with Euler Rotations
My solution:

    // Very similar to DrawModelEx but instead rotate on the 3 axis rather than defining one rotation
    void DrawModelAxisRotation(Model model, Vector3 position, Vector3 rotations, Vector3 scale, Color tint)
    {
        // Calculate transformation matrix from function parameters
        // Get transform matrix (rotation -> scale -> translation)
        Matrix matScale = MatrixScale(scale.x, scale.y, scale.z);
        Matrix matRotation = MatrixRotateXYZ(rotations);
        Matrix matTranslation = MatrixTranslate(position.x, position.y, position.z);

        Matrix matTransform = MatrixMultiply(MatrixMultiply(matScale, matRotation), matTranslation);

        // Combine model transformation matrix (model.transform) with matrix generated by function parameters (matTransform)
        model.transform = MatrixMultiply(model.transform, matTransform);

        for (int i = 0; i < model.meshCount; i++)
        {
            Color color = model.materials[model.meshMaterial[i]].maps[MATERIAL_MAP_DIFFUSE].color;

            Color colorTint = WHITE;
            colorTint.r = (unsigned char)((((float)color.r/255.0)*((float)tint.r/255.0))*255.0f);
            colorTint.g = (unsigned char)((((float)color.g/255.0)*((float)tint.g/255.0))*255.0f);
            colorTint.b = (unsigned char)((((float)color.b/255.0)*((float)tint.b/255.0))*255.0f);
            colorTint.a = (unsigned char)((((float)color.a/255.0)*((float)tint.a/255.0))*255.0f);

            model.materials[model.meshMaterial[i]].maps[MATERIAL_MAP_DIFFUSE].color = colorTint;
            DrawMesh(model.meshes[i], model.materials[model.meshMaterial[i]], model.transform);
            model.materials[model.meshMaterial[i]].maps[MATERIAL_MAP_DIFFUSE].color = color;
        }
    }

[TheEnbyWitch]

• Make the OBJ loader use the LoadTextData function for loading the .mtl files (or make the IO functions changeable). It was an issue when I wanted to embed every asset in the exe for a project, though I did end up using GLTF for the models (and its probably better size-wise).
• multiple windows! one of the features I'd love to have for the imgui implementation.

[raysan5]

• About OBJ loader, yes, that would be the desirable behaviour but it probably requires a review (or replacement) of tinyobjloader library.
• About multiple windows, there is an example in the Wiki to accomplish that... no plans to add that option to raylib for the moment...

[jmorel33]

As mentioned on the discord server, would be really nice to have union type variants of the Vector2 Vector3 and Vector4, which would allow more flexibility for our projects.

typedef union Vector2 {
    struct { float x, y; };
    float f[2];
} Vector2;

typedef union Vector3 {
    struct { float x, y, z; };
    struct { float r, g, b; };
    Vector2 xy;
    float f[3];
} Vector3;

typedef union Vector4 {
    struct { float x, y, z, w; };
    struct { float r, g, b, a; };
    struct { Vector2 xy, zw; };
    Vector3 xyz;
    Vector3 rgb;
    float f[4];
} Vector4;

This type of mapping is common in advanced scenarios.

Having this does not change the RayLib functionality or workflow, but permits us to access Vector types in a way more flexible manner.

Thanks

[raysan5]

I prefer to keep raylib structs simple.

[jmorel33]

I prefer to keep raylib structs simple.

I know you do I also like simple to understand structs. The benefit here is it allows to manmipulate the fields in a direct way, maybe it look slike obfuscating but in fact it is the opposite.

Anyhow not a big deal if you don't support but an impact on performance for my work and effort.

[brccabral]

I don't use the array notation, f[2], but particularly for Rectangles, IMHO this is a lot more easy to type, read, think about, and debug... And it is just one more line

typedef union Rectangle
{
    struct { Vector2 pos, size; }; // new line
    struct { float x, y, width, height; };
} Rectangle;

// example
Vector2 offset = {3, 4};
Rectangle rect = {0, 0, 5, 5};
Vector2 pos = Vector2Add(rect.pos, offset);
Rectangle newRect = {pos, rect.size};

[jmorel33]

Possibility to universalize the keyboard handling.

Way back when I was using Bliltz3D I solved this on the Windows platform simply by using: api_OemKeyScan% (wOemChar%) : "OemKeyScan"

https://docs.microsoft.com/en-us/windows/win32/api/winuser/nf-winuser-oemkeyscan

I made an example back then I shared with the community, as follows;

Dim scancodes(256)
For i = 0 To 255
   scancodes(i) = api_OemKeyScan(i) And $ff
Next
repeat_delay = 500
repeat_rate = 100
Print "TYPE HERE:"
While Not KeyDown(1)
   timer = MilliSecs()
   ascii% = GetKey()
   If ascii > 0 Then
      ascii_repeat = ascii
      scancode% = scancodes(ascii)
      repeat_timer = timer + repeat_delay
      Write Chr$(ascii)
   EndIf
   If KeyDown(scancode) = True Then
      If timer >= repeat_timer Then
         Write Chr$(ascii_repeat)
         repeat_timer = timer + repeat_rate
      EndIf
   Else
      scancode = 0
   EndIf
Wend

Essentially the keyboard scancodes could not be hardcoded and the OS would give them away, for keyboard keys.

[CreedVI]

Hey, just a tip for your formatting: to create a code block you use a back tick (`) not an apostrophe ('). The back tick key can be found above tab.

[jmorel33]

Hey, just a tip for your formatting: to create a code block you use a back tick (`) not an apostrophe ('). The back tick key can be found above tab.

Thanks

[raysan5]

At this moment keyboard handling goes through GLFW, I don't know how to integrate that option into raylib in a simple way.

[jmorel33]

At this moment keyboard handling goes through GLFW, I don't know how to integrate that option into raylib in a simple way.

I know what you mean. For windows I let the OS take care of the linguistics, and I just grab the standards scancodes.

[jestarray]

Support texture arrays?

[raysan5]

Out of scope for the moment. I think not many users would benefit of them.

[james-cabeludo]

Suggestion to add:

• GenMeshCapsule();
• CheckCollisionCapsuleModel(Mesh capsule, Model model);

https://wickedengine.net/2020/04/26/capsule-collision-detection/

Your work is amazing! Thanks for sharing this lib with us.

[raysan5]

Actually, GenMeshCapsule() could be probably accomplished using GenMeshCylinder() and GenMeshHemiSphere()... still it could be a nice addition.
About CheckCollisionCapsuleModel(), it's out of scope for raylib. There is no function for complex collision detection, I recommend using a physics library like ODE, Bullet or PhysX for advance collisions detection.

[sharkist]

• Add function DrawText with rotation parameter.
• consider to use WebGPU-native when will be final release instead of use Vulkan,Metal,DirectX in future.

[davidroman0O]

Just curious
Almost 2 years later
Still no plans for webgpu? :)

[raysan5]

no, out-of-scope for the moment

[davidroman0O]

Alright! Understandable

Thank you for your hard work and consistency 🙏

[gabrieldechichi]

Hey @raysan5 just curious: what are the reasons WebGPU is out of scope atm? And would that be something that could change in the future?

Also, would that be a feature an external contributor could feasibly implement, or is it too complex/requires major refactor so not feasible?

[raysan5]

It requires too much work. WebGPU support must be implemented as rlgl module, that is an abstraction layer

[jmorel33]

For the RayLib Windows gcc + npp to be offered as a 64bit package. Reason for this is for anyone that never played around with a compiler setup before, it can be puzzled as there is no how-to for this.

Also 32bit should not be supported anymore for Windows.as all x86 processors released since AMD64 are 64bit, that spans almost the last 20 years, which is twice as old as OpenGL 3.3.
https://en.wikipedia.org/wiki/X86-64

[zzador]

Just for your interest: You can allready build 64bit apps with the 32Bit GCC Compiler package. There is a difference between Compiler Host 64Bit and Compiler Output 64Bit. Long story short: The 32Bit Compiler can create 64Bit executables which can use the complete RAM however if your source-code gets bigger than 4GB the compiler might get problems compiling it cause it is just a 32bit compiler. If you want to create full 64bit code with the current GCC that is packaged with Raylib you just have to specify the target arch with the GCC parameter "-march=x86-64". Then your compiled code is fully 64Bit.

[raysan5]

Just note that despite computers support 64bit since 20 years ago, many of those computers still run 32bit operating systems. Actually, one of the main devices I use to develop raylib uses Windows 10 32bit. Probably in a future I'll update the installer to use MinGW 64bit... In any case, advance users could use the compiler they prefer.

[jmorel33]

Just for your interest: You can allready build 64bit apps with the 32Bit GCC Compiler package. There is a difference between Compiler Host 64Bit and Compiler Output 64Bit. Long story short: The 32Bit Compiler can create 64Bit executables which can use the complete RAM however if your source-code gets bigger than 4GB the compiler might get problems compiling it cause it is just a 32bit compiler. If you want to create full 64bit code with the current GCC that is packaged with Raylib you just have to specify the target arch with the GCC parameter "-march=x86-64". Then your compiled code is fully 64Bit.

Not sure I follow your 64bit explanation.

You might want to check the following thread and clarify so I can understand what you are writing here...
#1898

But essentially you're saying you don't need Mingw64 in order to compile 64bit. But somehow I was not able to compile 64bit prior to Mingw64. Once I switched to that version I easily managed to port everything to 64bit.

[DiiaBlood]

The literal only reason why I can't use Raylib is that it can't do Multi windows support and I know there is a way to modify the source code to allow multiple windows, I tried. I failed. Would absolutely love to see it in Raylib.

“Also transparent window click-through would also be cool ;)”

[raysan5]

I'm afraid multi-Window is out of scope for the moment, it's a specific use-case for the library and a work-around has been provided for advance users requiring it. Transparent Window mouse click-through could be considered.

[Peter0x44]

There is a branch that supports that here:
https://github.com/JeffM2501/raylib/tree/multi-monitor_50
The wiki page also has some info about it:
https://github.com/raysan5/raylib/wiki/Use-multiple-windows