eqrion
commented
1 year ago This sounds good to me. I just filed an issue here to discuss the future scope of the deterministic profile introduced in relaxed-simd. I would like to have that resolved before the phase 4 vote at the in-person meeting.
rossberg
After the recent CG meeting (TODO: link notes once uploaded), we need to pin down how we're advancing the intertwined Relaxed SIMD and Profiles proposals - we previously determined that Relaxed SIMD would have a dependency on the existence of a "deterministic" profile.
Here's my view on how votes for moving these forward should be structured:
We should vote on relaxed SIMD to phase 4 in the in-person meeting. This vote will imply editorial changes to the spec to create the syntactic framework for profiles and exactly two profiles ("full" and "deterministic") - the poll question can reflect this. No procedural mechanism will be established for adding any further profiles.
The profiles "proposal" should be interpreted as a consensus vote on establishing new procedure to evaluate and introduce future profiles - such as the suggested "no-gc" profile. Maybe this means that it shouldn't be called a proposal any more, but simply a vote on a procedural change - a more ambitious version of our previous phase advancement process changes. This debate+vote may take longer to work through (see https://github.com/WebAssembly/profiles/issues/6), and the relaxed SIMD vote above does not depend on it.
In the case that the first vote passes but the latter process never completes, this would result in "full" and "deterministic" indefinitely being the only profiles in the specification.