alan-baker

> > Much of the discussion in this bug seems to be about implementing inout using pointers. Is that the concern for WGSL? I thought it was implementing pointers using...

> so, the rules for pointer have that then pointer assignment is determined by compile time.. > > That makes implementation doable, but then I wonder what is the point...

This is done

The intention is that initializers are limited to creation-time expressions for module-scope variables; however, it appears that restriction was missed when creation-time expressions were introduced. Previously, the grammar enforced this...

Here are our opinions on the options: Pointers: 1. This strikes us as a scorched earth policy. Goes too far. There would be additional burden on the user to have...

> A question about something that was brought up in the meeting: it was stated that SPIR-V doesn’t have the ability to make a pointer that points to a field...

I get what you're saying. This is attempting to remind readers that you may have a "call stack" of in progress functions and that the stack may differ each time...

It feels like there's a missing clamp case for your examples: ```rust fn max_in__clamp2(from : u32, to : u32) -> u32 { let _arr_max = arrayLength(arr); let clamp_from = clamp(from,...

I think there are concerns about debugging the very non-local effects that could occur with trapping (e.g. vertex shader has OOB and the fragment shader writes 0s to the output)....

>Even if I do know the semantics of every attribute, I'll have easier time reading the code when I can skip parts of it quickly. I.e. if I see a...