raylib 3.0 roadmap

[raysan5]

Actually raylib 3.0 is just an extension to raylib 2.6 but with some big expected changes to the ABI that require lots of redesigns. Check #911 for reference.

Three big changes for this new release:

• Move all global variables to a global context defined by nested structs. For the moment, context will be defined by module to be self-contained and portable.
• Complete review and rewrite of logging system. Review logging system #1065
• Support custom memory allocators for raylib and all dependencies. Support custom memory allocators  #1074

Also considering:

• rnet module addition. [rnet] Simple and easy-to-use network module for raylib #753 -> Requires a full review and rewrite, not possible.

I'll update this issue with further info as required.

[hbiblia]

1 - It would be good to separate GLFW3 to easily implement other windows, such as SDL2, SOKOL_APP etc ...
2 - Separate the use of GLAD to be able to implement others.

[raysan5]

@hbiblia About your points:

1. No plans for that for the moment, you can check rayfork project for that approximation.
2. Actually I'd like to remove GLAD and just load the required extensions manually, raylib required OpenGL functionality is very limited, on OpenGL ES 2.0 I'm already doing that...

[chriscamacho]

have a look @ https://bitbucket.org/alfonse/glloadgen/wiki/Home
you can generate the core profile you want, then just cull all the stuff you don't need....

[brankoku]

If big changes are coming to the ABI, this would be the perfect opportunity to switch from pass by value to pass by reference. Syntax wouldn't change much, and it's more efficient to pass pointers around instead of filling the stack up with big structures.

[chriscamacho]

I agree with @brankoku especially for models, meshes etc...

[raysan5]

@brankoku sorry, no plans for that. I already did a big refactor to reduce all structures size, usually under 64 bytes and unify all structures to be used as pass-by-value (Music was a pointer reference before), I think it's easier for students and overload shouldn't be dramatic.

[raysan5]

Implemented global context state in commit 40b73a8.

[raysan5]

release publish checklist: #1084

[chriscamacho]

Just a thought, is it worth making the "context" structures non-static so more advanced users can make use of them ?

[pixelblender]

any plan for iOS support?

[raysan5]

@pixelblender no plans for the moment but you can check rayfork, it supports iOS.

[raysan5]

raylib 3.0 is complete.