Stream: rfc-notifications

Topic: rfcs / PR #37 Add support for the Component Model Async ABI

view this post on Zulip RFC notifications bot (Oct 08 2024 at 16:19):

dicej opened PR #37 from dicej:component-async to bytecodealliance:main:

Add support for the Component Model Async ABI to wasm-tools, wit-bindgen, and wasmtime.

Rendered RFC

view this post on Zulip RFC notifications bot (Oct 08 2024 at 16:35):

tschneidereit submitted PR review:

Hooray! :partying_face:

From a brief skim, the parts of this that I'm qualified to have an opinion on all look good to me. I left a couple of inline comments on specific aspects, but they're not particularly important.

view this post on Zulip RFC notifications bot (Oct 08 2024 at 16:35):

tschneidereit created PR review comment:

I think to some degree we don't really need this kind of discussion: the "what else could we do" question has been answered at the spec level, and we'd have had a ton of opportunity to voice disagreement there. At this point, I think we should handle this the same as other features, such as the recent support for Tail Calls.

As such, I wonder if there are possible alternative approaches to implementing support of the spec feature that would be worth discussing here? And if not, is there a way to explain why that is the case as part of a "there really isn't much of an alternative" paragraph?

view this post on Zulip RFC notifications bot (Oct 08 2024 at 16:35):

tschneidereit created PR review comment:

How (un)likely are scenarios in which all imports, but only some exports should be async, or vice-versa? If they might be reasonably likely, it might make sense to support All or None for one, and Some for the other. I assume you have a much better intuition on this than I do, though

view this post on Zulip RFC notifications bot (Oct 08 2024 at 16:36):

dicej updated PR #37.

view this post on Zulip RFC notifications bot (Oct 08 2024 at 17:13):

dicej submitted PR review.

view this post on Zulip RFC notifications bot (Oct 08 2024 at 17:13):

dicej created PR review comment:

Good point. If anything, I'd be tempted to remove the All option entirely, given that I expect most worlds will contain mostly non-blocking functions (e.g. wasi:http which contains just two functions which actually do I/O: wasi:http/types#[static]body.finish and wasi:http/handler#handle). And I suppose None is even more redundant; we could reduce it all to:

pub struct AsyncConfig {
  imports: HashSet<String>,
  exports: HashSet<String>,
}

Note that the proposed nonblocking function attribute will allow us to do the right thing automatically without any configuration required.

view this post on Zulip RFC notifications bot (Oct 08 2024 at 17:36):

dicej updated PR #37.

view this post on Zulip RFC notifications bot (Oct 08 2024 at 17:37):

dicej edited PR #37:

Add support for the Component Model Async ABI to wasm-tools, wit-bindgen, and wasmtime.

Rendered RFC

view this post on Zulip RFC notifications bot (Oct 08 2024 at 17:47):

dicej updated PR #37.

view this post on Zulip RFC notifications bot (Oct 08 2024 at 21:26):

dicej merged PR #37.

view this post on Zulip RFC notifications bot (Oct 08 2024 at 23:01):

dicej commented on PR #37:

Ugh, I definitely did _not_ mean to merge this. I must have pushed to origin/main by mistake. Sorry; will revert.

view this post on Zulip RFC notifications bot (Oct 08 2024 at 23:08):

dicej commented on PR #37:

I've reverted all the earlier commits, rebased, and opened https://github.com/bytecodealliance/rfcs/pull/38 -- please direct further discussion there. Sorry again for the confusion!

BTW, would it make sense to tweak the branch protection rules in this repo to block pushing straight to main? Asking for a friend.

view this post on Zulip Till Schneidereit (Oct 09 2024 at 21:00):

you can let your friend know that the main branch is now protected, as it should be

Last updated: Oct 23 2024 at 20:03 UTC