alexcrichton requested dicej for a review on PR #10844.
alexcrichton requested wasmtime-core-reviewers for a review on PR #10844.
alexcrichton opened PR #10844 from alexcrichton:backport-wit-bindgen-changes to bytecodealliance:main:
This commit wholesale copies the
wasmtime-wit-bindgenchanges from the
wasip3-prototyping repository to the Wasmtime repository here. This is
done to make the future merge withwasip3-prototypingsmaller
complexity-wise.This is intended to be a no-functional-change for non-"concurrent"
imports, or those for component model async. That means that it's
intended that no current embedding is affected by these changes.
Internally though there has been restructuring.Namely internally the generation of an interface trait, a resource
trait, and a world trait are all unified in a single method now. In the
future with component model async these traits are being split into
async/sync versions and so it was much nicer to do this in one location
rather than across three different locations.This required some minor renamings in the generated code and moving
around some impls, but otherwise the generated code should mostly be the
same as before (see golden output diff in the subsequent commit of this
PR)
dicej submitted PR review:
Probably out of scope for this PR, but I expect we could uncomment https://github.com/bytecodealliance/wasmtime/blob/d8b369b481cc1d291f6b7cb19700e53550116e92/crates/component-macro/tests/codegen.rs#L15-L23 and make it work by adding a stub
Accessorstruct and related items. That would give us _some_ test coverage for theconcurrent_imports: trueandconcurrent_exports: truecases.BTW, I just noticed that code is commented-out in the
wasip3-prototypingrepo also. Probably an accidental artifact of merging?
alexcrichton commented on PR #10844:
Oops yeah that's an accident, and yeah I'll work on uncommenting this in wasip3-prototyping and then see what it would take to merge back those changes here
alexcrichton merged PR #10844.
Last updated: Dec 06 2025 at 06:05 UTC