Loopback service stubs accept a dynamic version (cohort) parameter

[sohpeach]

Similar to props, this allows a worker to call the service stub as a function, passing along the given version information (currently only a cohort string) to the runtime.

workerd's server implementation currently ignores this information, so the test included in this commit is very minimal. We could come up with a better test once we've decided how this information will be exposed to the worker, assuming it also makes sense in the workerd runtime. The internal codebase tests this feature in more depth.

The feature is behind the experimental enable_version_api flag. Usage looks like:
let service = ctx.exports.default({ version: { cohort: "blah" } });

[github-actions]

The generated output of @cloudflare/workers-types matches the snapshot in types/generated-snapshot 🎉

[codspeed-hq]

Merging this PR will improve performance by 19.42%

⚡ 1 improved benchmark
✅ 69 untouched benchmarks
⏩ 129 skipped benchmarks¹

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
⚡	jsonResponse[Response]	47.5 µs	39.8 µs	+19.42%

---

_{Comparing swallace/loopback-service-binding-cohorts (4fa1ce8) with main (8719dcd)}

Footnotes

1. 129 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[ask-bonk]

@sohpeach Bonk workflow was cancelled.

View workflow run · To retry, trigger Bonk again.

[sohpeach]

/bonk review

[ask-bonk]

Review posted. Here's a summary of the findings:

Review Summary

I reviewed PR #6152 which adds a dynamic version (cohort) parameter to loopback service stubs.

Findings posted:

1. [LOW] Unused include -- #include <workerd/io/features.h> in export-loopback.c++ is not used. Likely left over from an earlier iteration.

2. [LOW] versionRequest silently ignored in server.c++ -- The new parameter is accepted but never read. A // TODO comment would help future readers understand this is intentional (the PR description explains it, but code outlives PR descriptions).

3. [LOW] Test coverage is minimal -- The test confirms the new API shape doesn't crash, but doesn't verify version propagation. jp4a50's earlier feedback about also passing props and testing undefined appears unaddressed.

Non-issues confirmed:

• jsg::Optional<kj::Maybe<kj::String>> cohort correctly models the JS undefined | null | string semantics
• The two-method approach (call vs callWithVersion gated on compat flag) preserves backward compat
• VersionRequest::clone() is correctly implemented
• EW_EXPORT_LOOPBACK_ISOLATE_TYPES correctly includes the new struct types
• TypeScript snapshots match the C++ definitions

Note: I was unable to post inline comments because the remote PR head appears to have been overwritten with a different commit. The overall review comment was posted successfully.

---

Time for a pun! I tried to review a cohort parameter, but it just kept versioning away from me.

github run

[jasnell]

Time for a pun! I tried to review a cohort parameter, but it just kept versioning away from me.

Ha!! Bonk picked up on the dad-jokes skill I added. Love it.