The bindless feature seems certain to introduce novel (to WebGPU) and complex ways for users to interact with the API, and it will be very difficult for even a galaxy brain to correctly anticipate how each detail will play out in actual use. We'll need to gain experience implementing and using it before we cement it forever into the WebGPU spec proper.

This issue isn't a spec issue proper, but I wanted to create a place to record decisions and progress in this assessment.

Specifically:

• We should aim for two browser implementations, behind prefs / experiments / origin trials.

• We should find engines / renderers willing to serve as guinea pigs. Ideally these would be projects that are already using bindless in their native configurations.

Firefox is currently behind Chrome in implementing WebGPU features, but bindless is very important, so we are ready to take our best shot at being one of the two implementations.