bevy-cheatbook VS wgpu-rs

You can see working examples of nearly everything you asked about in the Bevy Cheatbook

The biggest source of wasted time for me was finding out that my shaders which were working on native didn’t work on webgl. For one of the more perplexing error messages, I opened this issue in the cheatbook a while ago.

wasm mostly works :) There's a nice PR open on the cheatbook that documents the process quite well that you can work off of.

Great! Are you okay if I submit this to the Unofficial Bevy Cheatbook? This question comes up pretty frequently and this is a great little snippet.

Naga-based shader infrastructure has been growing and capturing more ground. It has reached an important point where SPIRV-Cross is not just optional on some platforms, but even not enabled by default. This is now the case for Metal and OpenGL backends. Naga path is easier to integrate, share types with, compile, and it’s much faster to run. Early benchmarks suggest about 2.5x perf improvement over SPIRV-Cross for us.

wgpu-rs (https://github.com/gfx-rs/wgpu-rs) is probably closer to what you're looking for, it's a "high level wrapper" over wgpu-core (Which implements "WebGPU", an upcoming browser API for graphics using gfx-rs, and AIUI "implements" means this is what a browser might use to actually call a graphics API when javascript uses webGPU, although if you're compiling for wasm it could just call the browser APIs). It can run natively, ignore the fact it has "Web" in the name

rust-gpu) is used for compiling a compute shader written in Rust to SPIR-V. wgpu-rs is used natively for running the GPU computation.

For web assembly there is the unofficial bevy_webgl2 plugin. Official bevy web assembly support would probably use the wgpu webgl backend, which still needs some work, and is currently untested in bevy.

We let wgpu https://github.com/gfx-rs/wgpu-rs handle our graphics backend abstraction, so our OpenGL support will come whenever they implement and release it. Currently, it seems to be a WIP.

Same as with GLSL, isn't it? What we do in wgpu-rs examples, and that's something I expect to see more widely, is having an integration test that just parses all the WGSL in the project and reports errors. Ideally though, we'd have a set of IDE plugins to do the parsing and report errors right where you type the code. We'd appreciate any help to get these started!

Also you could use an existing cross-platform graphics library, e.g. wgpu-rs. It supports Vulkan, Metal, DirectX, OpenGL and WebGPU. Oh and it seems to allow switching backends at compile-time via the WGPU_BACKEND environment variable, so maybe check out how they're doing it.

there is a lack of opengl development because opengl itself is beint phased out slowly. your best bet here is probably https://wgpu.rs