Phase 4: validate Rust indexing readiness on VS Code or a same-class large JS/TS repo

[jununfly]

This was generated by AI during triage.

What to build

Add large-repository readiness evidence for the opt-in Rust JS/TS indexing path. VS Code is the default target; a same-class large JavaScript/TypeScript repository may be substituted only with a documented reason and equivalent profile/sufficiency evidence.

Acceptance criteria

• A VS Code commit is pinned, or a same-class replacement is documented with repository URL, commit, reason, and indexed file count.
• TypeScript and Rust indexing profiles are run on the large target.
• Results record wall-clock, peak RSS, finalization subphases, node/edge counts, and dominant bottleneck.
• At least one Explore sufficiency probe is run against the large target.
• Rust indexing does not increase generic Read/Grep fallback risk in the large-target sufficiency probe.
• Raw artifacts and a compact summary are stored for the Phase 4 results document.
• The large target remains outside the ordinary quick local test loop unless an explicit long-running validation command is used.

Blocked by

• Phase 4: make Rust indexing RSS and profile evidence repeatable #78

[jununfly]

This was generated by AI during triage.

#81 status

Large-target evidence is now collected and documented.

Summary:

• Target: microsoft/vscode at 275e1b3102b410cb6fe8a8dd7d12bc5cebb917f0.
• Target shape: large VS Code JS/TS sparse checkout, not full VS Code.
• Indexed JS/TS files: 11,291.
• TypeScript profile: 224.8s, peak RSS 1.30GB.
• Rust opt-in profile: 256.7s, peak RSS 1.46GB.
• Rust node/edge counts: 557,770 / 1,648,219.
• Dominant bottleneck: referenceResolutionMs.
• Explore sufficiency probe: TypeScript and Rust both returned connected Flow sections with 0 / 0 deterministic Read/Grep fallback-risk signals.

Artifacts:

• docs/benchmarks/2026-06-13-rust-indexing-core-phase-4-large-target-readiness.md
• docs/benchmarks/2026-06-13-rust-indexing-core-phase-4-vscode-profile.raw.json
• docs/benchmarks/2026-06-13-rust-indexing-core-phase-4-vscode-sufficiency.raw.json
• docs/benchmarks/2026-06-13-rust-indexing-core-phase-4-vscode-sufficiency-prompts.json

Decision consequence:

#81 is complete as an evidence issue, but the result does not support default rollout. The Phase 4 path remains continue opt-in plus targeted blockers.

Follow-up blockers have been split out:

• Phase 4: rerun VS Code readiness smoke under supported Node #85
• Phase 4: classify VS Code large-target parse errors #86
• Phase 4: investigate large-repo reference resolution bottleneck #87