[robobun]

Problem

The parallel-batch allowlist (test/parallel-allowlist.json) was generated from builds 82146-83745, which predate #36175 (the PR that introduced the parallel batch). The generator's rule is "zero flaky/failed annotations in the scanned builds", but at that point no test had ever run inside a shared bun test --parallel process, so the allowlist couldn't know which tests are incompatible with in-process parallelism.

The result: tests that measure process-wide state (heapStats().objectTypeCounts, rss(), expectMaxObjectTypeCount) or have tight wall-clock budgets were included, and they now flake on nearly every build. Across the last 400 completed PR builds, the top "failed in the parallel batch; passed alone" annotations by distinct build count:

test	builds (of 400)
test/js/web/streams/streams-leak.test.ts	348
test/js/bun/shell/leak.test.ts	340
test/js/bun/net/socket-retention.test.ts	334
test/js/web/fetch/fetch-tls-abortsignal-timeout.test.ts	282
test/js/bun/spawn/spawn-streaming-stdout.test.ts	267
test/regression/issue/08965/08965.test.ts	261
test/regression/issue/20875.test.ts	151
test/js/node/http/node-http-pinned-write.test.ts	110
test/regression/issue/27272.test.ts	103
test/js/third_party/esbuild/esbuild-child_process.test.ts	103

Each of these generates a yellow annotation + serial retry on most builds.

Fix

Re-ran scripts/update-parallel-allowlist.mjs (300 PR builds, 85133-86444). 609 distinct files flaked or failed in that window; the generator drops them per its existing rule.

• 282 dirs -> 215
• 1642 files -> 1296 eligible for the batch
• 153 -> 160 explicit excludes

All six of the top offenders above are now excluded (either via excludeFiles or their dir dropped).

Verification

streams-leak.test.ts asserts rss() < 250MB; in the batch it sees 390MB-2.6GB from neighbouring tests. shell/leak.test.ts's #11816 tests assert heapStats().objectTypeCounts.ShellInterpreter <= 3 in-process; any concurrent test using $ contributes. socket-retention.test.ts and spawn-streaming-stdout.test.ts use expectMaxObjectTypeCount similarly. These are correct tests in an isolated process and cannot be made parallel-safe without moving the assertions into a subprocess; excluding them from the batch is what the generator already does once it has data.

---

no test proof · iteration 0 · docs-only change; test-proof not applicable

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro

Run ID: 879b9d50-609e-4f2f-94a7-6294d4567fd8

📥 Commits

Reviewing files that changed from the base of the PR and between f68e504 and 08b90a6.

📒 Files selected for processing (1)

• test/parallel-allowlist.json

---

Walkthrough

The generated parallel-test configuration now contains updated build metadata, directory allowlists, and excluded-test files across Bun, Node.js, third-party, web, and regression suites.

Changes

Parallel test allowlist

Layer / File(s)	Summary
Regeneration metadata and root coverage test/parallel-allowlist.json	The generation metadata, scanned build range, aggregate counts, and initial directory coverage were updated.
Parallel directory coverage test/parallel-allowlist.json	The allowlist was recalculated across Bun, Deno, first-party, Node.js, third-party, web, N-API, and regression test directories.
Excluded-test inventory test/parallel-allowlist.json	The excluded-test list was replaced with a new inventory covering multiple test suites.

Possibly related PRs

• oven-sh/bun#35944: The regenerated exclusions may include parallel-safe test exclusions addressed by this PR.

Suggested reviewers: jarred-sumner

🚥 Pre-merge checks | ✅ 4

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly states that the parallel allowlist was regenerated from post-#36175 build data.
Description check	✅ Passed	The description explains the problem, fix, data changes, affected tests, and verification evidence.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

---

_{Comment @coderabbitai help to get the list of available commands.}

[claude]

LGTM — mechanical regeneration of the parallel-batch allowlist from newer CI data.

What was reviewed:

• Consumer in scripts/runner.node.mjs:388-394 — the new "" dir entry is handled correctly (matches top-level package-json-lint.test.ts only).
• Self-test test/internal/parallel-allowlist.test.ts invariants: all 160 excludeFiles exist on disk and have their dirname in dirs; counts match _meta.stats (215/160).
• All ten top-offender tests from the PR description are excluded (either in excludeFiles or their dir dropped).

Extended reasoning...

Overview

Single-file change to test/parallel-allowlist.json, a generated CI configuration file that tells scripts/runner.node.mjs which test files may run in the shared bun test --parallel batch. The generator script (scripts/update-parallel-allowlist.mjs) is unchanged — only its output is refreshed against builds 85133-86444, which now include real parallel-batch flake data (the previous window predated the batch's introduction in #36175).

Net effect: 282→215 dirs, 153→160 explicit excludes, 1642→1296 batch-eligible files. Tests dropped from the batch fall back to serial execution — no coverage is lost.

Security risks

None. This is CI test-scheduling metadata; it does not touch runtime, build, or any shipped code path.

Level of scrutiny

Low. The failure mode of a wrong entry is a yellow flake annotation + serial retry (the exact symptom this PR is fixing), not a broken build or missed test. I verified structural correctness against both the runner's consumer logic and the repo's own self-test:

• jq counts match _meta.stats exactly (215 dirs, 160 excludes).
• Every excludeFiles entry exists on disk and its dirname() is present in dirs (so test/internal/parallel-allowlist.test.ts passes).
• The new "" dir entry is intentional: isParallelAllowlisted uses slash === -1 ? "" : path.slice(0, slash), so it matches only top-level test files — currently just package-json-lint.test.ts.
• Spot-checked all ten top-offender tests from the PR table: each is now excluded either via excludeFiles (streams-leak, shell/leak, socket-retention, fetch-tls-abortsignal-timeout, 20875, node-http-pinned-write, 27272) or by dropping their dir (js/bun/spawn, regression/issue/08965, js/third_party/esbuild).

Other factors

The PR description's mechanism analysis (RSS/heapStats assertions being process-wide) is sound and matches why these tests can't be made parallel-safe without subprocess isolation. This is exactly the kind of periodic data refresh the generator was built for.

[github-actions]

Found 3 issues this PR may fix:

1. node-net.test.ts: un-awaited expectMaxObjectTypeCount in named-pipe test fails unrelated subsequent tests on Windows #32092 - PR removes js/node/net from the parallel-safe dirs, preventing the un-awaited expectMaxObjectTypeCount in node-net.test.ts from contaminating subsequent tests in the batch
2. ASAN CI: JSC assertion in JSObject::getOwnPropertyDescriptor during worker terminate (test-worker-message-port-transfer-terminate) #34095 - PR removes js/node/worker_threads from the parallel-safe dirs, reducing the concurrency that triggers the intermittent JSC assertion failure during worker terminate
3. ASAN CI: ExceptionScope::assertNoException during worker terminate (worker-transfer-terminate-stress, separate from #34095) #34690 - Same as above — removing js/node/worker_threads from parallel batches mitigates the timing-sensitive ASAN ExceptionScope::assertNoException race during worker termination

If this is helpful, copy the block below into the PR description to auto-close these issues on merge.

Fixes #32092
Fixes #34095
Fixes #34690

🤖 Generated with Claude Code

[robobun]

Re the find-issues bot: none of #32092, #34095, #34690 are fixed by this PR.

• ASAN CI: JSC assertion in JSObject::getOwnPropertyDescriptor during worker terminate (test-worker-message-port-transfer-terminate) #34095 / ASAN CI: ExceptionScope::assertNoException during worker terminate (worker-transfer-terminate-stress, separate from #34095) #34690 are the ExceptionScope::assertNoException abort during worker terminate. worker-transfer-terminate-stress.test.ts was never in the parallel batch (isBucketCandidate excludes /stress/i), and test-worker-message-port-transfer-terminate.js is a node-style test (also excluded). The 44 hard-red hits I counted are serial-run failures; they need the WebKit fix in JSModuleLoader: don't treat a TerminationException from resolve() as a resolution failure WebKit#309 / JSModuleLoader: do not abort when terminate() lands during a worker’s module loads (WebKit pin bump) #34655.
• node-net.test.ts: un-awaited expectMaxObjectTypeCount in named-pipe test fails unrelated subsequent tests on Windows #32092 is a missing await inside node-net.test.ts; the floating promise contaminates later tests in the same file run regardless of whether that file is in the batch.

Leaving the PR description as-is.

[robobun]

Heads up: the regen dropped js/bun/util/inspect-error-leak.test.js from excludeFiles because the scanned window (85133-86444) only contained builds where it was already excluded, so the generator saw zero flakes. Once included it times out on x64-asan and crashes the parallel worker on Windows on most builds since 86456. Re-added in #36867.