Dzmitry Malyshau

Github suggests me to review, but I don't really have any energy to maintain collision-rs. If nobody shows up, we can call this abandonware, unfortunately.

Thanks for filing this! 1. I think it's acceptable for MVP, it's not the worst place to be in. 2. This would really help to bring the overhead to near-zero...

Sounds reasonable (explanation and last suggestion)

@litherum thank you, great material here! > So, a bindless design can be achieved by building an argument buffer with all necessary resources, and making sure that the necessary ones...

> basically descriptor-indexing is a better deisgn for an implementation that has "performing validation and verification" as one of the concerns in its scope. I think this can be done...

> Both these fall into the (2) > Do you want to change that? I still stand by my past comment *in principle*: > separate two types of accesses statically:...

@dneto0 we talked about this internally, and I think there is a problem with the consensus. If `[]` syntax is only for dynamic indexing, then by-value arrays become non-indexable. Previously,...

@dneto0 sorry, I think you misunderstood the concern, which I explained poorly. Let me try again. So we appear to talk about 2 different notions of "dynamic access", that is...

Concrete example that tripped one of the workflows for us: ```rust let _e21: mat4x4 = global1.Model; mat3x3(_e21[0].xyz, _e21[1].xyz, _e21[2].xyz) ``` If `[]` is considered "dynamic access", then this code is...

I don't think the matrix access by letter is a solution here. It allows "static" access into matrices, but there are still fixed-size arrays left on the table. If I...