Stream: git-wasmtime

Topic: wasmtime / PR #15054 CM+GC: support arbitrarily nested tu...

view this post on Zulip Wasmtime GitHub notifications bot (Oct 28 2026 at 22:19):

fitzgen opened PR #15054 from fitzgen:cm-gc-nested-tuples to bytecodealliance:main:

Depends on https://github.com/bytecodealliance/wasmtime/pull/15051

view this post on Zulip Wasmtime GitHub notifications bot (Oct 28 2026 at 22:19):

fitzgen removed the review request for wasmtime-core-reviewers from PR #15054.

view this post on Zulip Wasmtime GitHub notifications bot (Oct 28 2026 at 22:19):

fitzgen requested pchickey for a review on PR #15054.

view this post on Zulip Wasmtime GitHub notifications bot (Oct 28 2026 at 22:19):

fitzgen requested wasmtime-core-reviewers for a review on PR #15054.

view this post on Zulip Wasmtime GitHub notifications bot (Oct 28 2026 at 22:51):

pchickey removed the review request for pchickey from PR #15054.

view this post on Zulip Wasmtime GitHub notifications bot (Oct 28 2026 at 22:51):

pchickey requested alexcrichton for a review on PR #15054.

view this post on Zulip Wasmtime GitHub notifications bot (Oct 29 2026 at 20:06):

:memo: alexcrichton submitted PR review:

Some of this is going to be dependent on the PR this is built-on, but my general impression here is that this is a fair bit of special-casing which naively seems like it may not be necessary with some refactoring in other places. Would it be possible to experiment to shift things around to do that? For example how come HelperLocation::Stack can't be used for GC references since it's always a single param and a single result?

view this post on Zulip Wasmtime GitHub notifications bot (Oct 29 2026 at 20:06):

:speech_balloon: alexcrichton created PR review comment:

In the interest of avoiding special-casing where possible, could this be removed? Similarly below the special-case to skip the get/set on the stack seems like it could be removed too. I'd otherwise expect that a GC reference is basically always represented as "stack" since that's basically another name for "in a local"

view this post on Zulip Wasmtime GitHub notifications bot (Oct 29 2026 at 20:24):

fitzgen commented on PR #15054:

For example how come HelperLocation::Stack can't be used for GC references since it's always a single param and a single result?

See https://github.com/bytecodealliance/wasmtime/pull/15051#discussion_r4417440249

Would it be possible to experiment to shift things around to do that?

Sure, I can poke at some more stuff, but if you have any other ideas, let me know...

Last updated: Oct 31 2026 at 23:03 UTC