Note that these are not examples, anything in src/Lab often demonstrates terrible coding practice and is likely extremely bug prone, so it’s probably better if you just pretend those don’t exist.

I don't think it matters in this context. a 300 lines webgpu sample that can be refactored by OP is better that not knowing where to start at all. I am in fact thankful for Lab folder because it has a lot of good use cases for Silk.NET , otherwise Silk is unusable for beginners if your logic is applied.

Also it is partially true. for example. in sample of ImGui and Vulkan, the code can be refactored but it is perfectly functional and runs on Mac/Windows/Linux and helped me bootstrap a good gpu driven rendering with C# and Silk.NET

I don't think it matters in this context. a 300 lines webgpu sample that can be refactored by OP is better that not knowing where to start at all. I am in fact thankful for Lab folder because it has a lot of good use cases for Silk.NET , otherwise Silk is unusable for beginners if your logic is applied.

the WebGPU code in the lab is very bad, i can attest as i am the one who wrote it, it forgets to delete a bunch of stuff and shows sub-optimal use of the API, id reccomend looking at the wgpu-native samples to see how to use the API, its not in C#, but our bindings are 1:1, so it should be about the same
https://github.com/gfx-rs/wgpu-native/tree/trunk/examples

the webgpu lab code is basically just a "smoke test" for the bindings