[kixelated]

Problem

The ts differential tests run each fixture's native wasm32 side as a debug build, which V8 runs several times slower, with V8's background threads adding CPU on top. stable_names also translated lang a second time as its base, with the same roots as the lang test's own translation. Per the 2026-11-01 checks audit, that made lang plus stable_names most of just ts test's CPU.

Approach

• A differential profile in ts/Cargo.toml inherits dev with opt-level = 1. Overflow checks and debug assertions stay on, so panics and overflow match exactly. Every fixture's native side now builds with it, and stock's getrandom override moves to that profile.
• Node runs differential.ts with --single-threaded.
• lang's translation sits in a OnceLock that lang and stable_names share. stable_names checks its added-root translation against that output instead of translating a base of its own. added_root splits into rooted (translate with replaced roots) and kept (the line-order check), so source_order still translates its own base.

Evidence

Every fixture compares the same cases, with the same panic and abort counts as origin/main and 0 differences (--nocapture, cold target/tmp):

fixture	cases	panics	aborts	differences
lang	709442	84411	96	0
task	35417	3059	0	0
codec	35273	0	0	0
stock	20279	2296	0	0
bind, crates, and share's 4 builds	16273 each	1387 in one, 1 in another, 0 in the rest	0	0

Breaking the runtime by one (u32Add's bound at 0xfffffffe) fails lang: 709442 cases, 84411 panics, 96 aborts, 13 differences, e.g. native ffffffff01, ts panic "attempt to add with overflow".

Cold cargo test --locked in ts/ (test binaries prebuilt, target/tmp removed), measured on this host (32 cores, shared with other agents, so wall time is noisy; CPU is the measure):

	CPU (user + sys)	wall	lang	stable_names
origin/main, run 1	2247 s	9m37s	576 s	341 s
origin/main, run 2 (load ~50)	2805 s	12m13s	729 s	397 s
this PR, run 1	1195 s	8m04s	479 s	348 s
this PR, run 2 (load ~50)	1194 s	8m54s	506 s	353 s

About 1050 to 1600 CPU-seconds saved per cold run (47% to 57%). Wall time is bound by lang's translation, which this PR does not change.

Impact

• Public API: none. Test infrastructure only.

Alternatives

• --config profile.dev.opt-level=1 on the build command: hides the profile from Cargo.toml and shares debug/ with any other dev build of the fixtures.
• Merging stable_names into the lang test: shares the translation without a OnceLock, but serializes its added-root translation after lang's differential and loses its separate test name.
• debug = false in the profile as well: not measured, left out to keep the change to what the audit measured.

Follow-ups

• Review round 1's finding that borrows::leaks was unreachable came from a base before fix(ts): hold a byte vector's capacity as a heap or deque takes it #417. fix(ts): hold a byte vector's capacity as a heap or deque takes it #417 picks a case by two bytes, so borrows::leaks is [0x00, 0xff], and op refuses any case a 0xff first byte would shadow. The quest planned for it in this branch is dropped again, and nothing new is planned.
• Stable names on a small fixture still applies: it drops stable_names' remaining lang translation, though it can no longer share lang's.

Checks

just ci passed in a fresh nix develop on head bfbd5dc, after merging origin/main (345aa82). Running the built native lang wasm confirms [0x00, 0xff, ...] runs borrows::leaks, not Fault.

Review

Codex gpt-6.1-sol, medium effort, via quest review --base origin/main.

• Round 1 (head 10a3600), 2 findings, both outside this diff (origin/main moved to b1656c9 during the session):
  • ts/runtime/rt.ts:1381 byte-vector capacity on heap and deque conversion: refuted, already fixed on origin/main by fix(ts): hold a byte vector's capacity as a heap or deque takes it #417 (byteVecToVec uses capHold).
  • ts/fixtures/lang/src/harness.rs:300 case 255 unreachable: refuted. The base predated fix(ts): hold a byte vector's capacity as a heap or deque takes it #417's two-byte selector, so on origin/main borrows::leaks is [0x00, 0xff] and runs. The quest first recorded for it is deleted.
• Round 2 (head ef428a9): no findings.
• Head bfbd5dc: review skipped. Since ef428a9 the branch only has a conflict-free merge of origin/main and the deletion of the moot quest quest/m0/lang-cases.md, with no source change.

🤖 Generated with Claude Code

[coderabbitai]

Important

• Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

Run configuration

• Source mode: Auto
• Repository: Unavailable
• Branch: Unavailable
• File: No configuration file from this repository applied
• Commit: Unavailable
• Layers: Organization UI
• Review profile: CHILL
• Plan: Advanced
• Run ID: 394a7168-342b-4896-8686-212bd94f06ee

No actionable comments were generated in the recent review.

Recent review info

Run configuration

• Source mode: Auto
• Repository: Unavailable
• Branch: Unavailable
• File: No configuration file from this repository applied
• Commit: Unavailable
• Layers: Organization UI
• Review profile: CHILL
• Plan: Advanced
• Run ID: 83917acb-0a21-4aaf-9e35-8b07306cd5e8

Commits

Reviewing files that changed from the base of the PR and between 345aa82 and bfbd5dc.

Files selected for processing: 5

• quest/m0/checks/README.md
• quest/m0/checks/differential.md
• quest/m0/checks/stable-names.md
• ts/Cargo.toml
• ts/tests/fixtures.rs

Files with no reviewable changes: 1

• quest/m0/checks/differential.md

Included review availability: This review used your included allowance. Your plan provides up to 1 included review per hour; 0 remain after this review.

Walkthrough

The TypeScript differential fixture harness now uses an optimized Cargo profile and runs Node with --single-threaded. Root-order checks compare outputs from separate base and added-root translations. The lang translation is cached for reuse, and the stable-names plan reflects this change.

Merge Risk: Minimal · previous review bfbd5

No actionable merge-blocking issue is established for these test-infrastructure changes. Merge after normal checks pass.

Pre-merge checks: 5

Check name	Status	Explanation
Title check		The title names the two main changes: the differential fixtures now run optimized and single-threaded, and lang's translation is shared. Both match the diff: a new differential Cargo profile with opt-…
Description check		The description states the problem, the approach, benchmark evidence, impact, alternatives, follow-ups, and checks. It matches the changeset: the differential profile in ts/Cargo.toml, the --single-th…
Docstring Coverage		Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff with cyclomatic complexity of at least 2. Documentat…
Linked Issues check		Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check		Check skipped because no linked issues were found for this pull request.

Full details: Title check

Explanation

The title names the two main changes: the differential fixtures now run optimized and single-threaded, and lang's translation is shared. Both match the diff: a new differential Cargo profile with opt-level 1, Node running with --single-threaded, and a OnceLock shared between lang and stable_names. The title is specific and concise, though it is long (100 characters) and uses the chore(ts) prefix.

Full details: Description check

Explanation

The description states the problem, the approach, benchmark evidence, impact, alternatives, follow-ups, and checks. It matches the changeset: the differential profile in ts/Cargo.toml, the --single-threaded Node flag, the OnceLock shared translation, and the split of added_root into rooted and kept. It also covers the removed quest documents.

Full details: Docstring Coverage

Explanation

Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff with cyclomatic complexity of at least 2. Documentation-only edits and deletions within surviving functions are included; unchanged diff context and fully deleted functions are excluded. Analyzed 1 functions across 1 files. (2 skipped: 2 unsupported.)

Finishing Touches

Simplify code

• Commit to this branch
• Create a new PR

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

Share

• X
• Mastodon
• Reddit
• LinkedIn