Kelsey Gilbert

Specifically, I believe it was AB who feels that we're probably satisfactory here, and that MM will check our current spec text to confirm.

WGSL meeting minutes 2022-05-24 * MM: Think there is group consensus to restrict some kinds of aliasing, but how far do we go? As I understand, SPIR-V says no-aliasing anywhere...

WGSL meeting minutes 2022-05-31 * MM: Couple thoughts, homework was to read spec and figure out what spec said. Little interested in the statement that says it is a dynamic...

WGSL meeting minutes 2022-06-07 * MM: Internal discussion raised: * If compiler can prove that two pointers don’t alias, then in some programs the compiler can generate better code around...

WGSL meeting minutes 2022-06-14 * MM: Talked internally. Our general position is unchanged. We do understand that in HLSL, for in-out params, it’s not generally possible to support aliasing, to...

WGSL meeting minutes 2022-06-21 * MM: Seem to remember close to agreement. * KG: DN wanted to think about it * DN: Stuck on the other issue. Concerned we get...

WGSL meeting minutes 2022-06-28 * MM: Interesting comment about how HLSL defines that it always uses copyin/copy out even on device with pointers it’s not allowed to use them. Followup...

WGSL meeting minutes 2022-07-05 * MM: Posted examples * AB: The first one: the compiler has to see the aliasing of y and z to x. So myBuffer[0] gets 4...

WGSL meeting minutes 2022-07-26 * Previously * JB: Let’s talk, I want time to think on this. Can we come back next week? (yes) * JB: Lots of discussion in...

WGSL meeting minutes 2022-08-02 * JB: I believe that most of the topics raised by MM are largely solved/handled/decided. The big thing left is whether aliasing is a static or...