[devin-ai-integration]

What and why

The 0.11.0 performance record that merged with #1337 measured an earlier candidate (develop at d8b60ed62 with #1350 merged on top) and described develop as not having moved; by the time it merged, every fix it named had merged too and develop had advanced by 23 further merges. This rewrites docs/project/performance-release-0.11.0-vs-0.10.1.md as a measurement of the head the release is cut from, 5db0f74158f1ecd43509d90141e5055cb0c3e539, against the v0.10.1 tag:

• six fresh rounds of every benchmark package on that head against the existing six rounds of v0.10.1, benchstat over both; one candidate column instead of three;
• a quiet interleaved pass (0.10.1 | candidate, six rounds) over internal/exec/runtime, internal/frontend/{repl,lsp,grpc}, internal/workspace/model, the tests/perf rows whose loaded time had moved and the 32-satellite stress rows, which is every time in the tables not marked †;
• whole-binary validation, start-up floor and binary sizes re-taken on the shipping binaries (parity on time at every size; resident set +3–7 MiB; sysml 47.60 → 49.32 MiB);
• every movement re-attributed against the new head: the executor is at 0.10.1 to the allocation after perf(runtime): compute an expression's footprint once per model #1347/perf(semantic): compute a feature's redefinition closure once per model #1353/perf(runtime): compute a statement's order and stamped footprint once per model #1355/perf(runtime): take a state machine's transition without allocating #1356; the analysis rows carry fix(runtime): type an object by every type its feature inherits through subsetting #1276/feat(semantics): usages subset the base library usage of their kind #1319 at +4–18% (the earlier candidate: +27–89%); RenameWarm now allocates less than 0.10.1 and ReferencesCold 2.8× more (perf(resolve): keep the members an import surfaces for the query asking again #1358's first enumeration per import, a named design cost); DecodeSnapshot +6.8% is perf(semantics): read a library symbol's implied generals from its record #1357's recorded implied generals;
• a regression no earlier candidate had, found by this measurement: Instantiate 2.51 M → 6.34 M allocations, BatchSatisfy +93%, FeaturesOf +56%, GRPCVerifyConstraint +50%, bisected to fix(runtime): nested action and state usages are performances their owner holds #1310 and fix(runtime): bind a nested requirement usage's subject from the owning object #1312. The avoidable parts are fixed in perf(semantics): decide a member's root restatement once per symbol #1368 (semantics: the root-restatement walk memoized per symbol) and perf(runtime): read a declaration's member from its scope's index #1369 (runtime: a declaration's member read from the scope's index), whose six quiet rounds are the record's third column where it appears (with both, Instantiate 2.22 M, below 0.10.1); GRPCVerifyConstraint's +4.9 k is fix(runtime): nested action and state usages are performances their owner holds #1310's own work and is carried.

The record also states what the quiet pass leaves unexplained: four executor rows with identical allocation counts that moved +16–24% on time at p = 0.009–0.026, carried as time unexplained by allocation rather than attributed. The changelog fragment's wording is adjusted to name the head measured; mkdocs.yml, docs/project/README.md and docs/internals/performance.md already link the record and are unchanged. No implementation code is in this PR.

How it was verified

• python3 scripts/check-doc-ids.py, python3 scripts/changelog.py check, python -m mkdocs build --strict, go test ./tests/hygiene clean. python3 scripts/check-doc-links.py reports one broken link in docs/project/third-party-notices.md (../assets/landing/libavoid-js.LICENSE.txt), which is the same on develop and not touched here.
• Raw rounds, benchstat tables and the whole-binary runs are reproducible from the method section; the two fix PRs carry their own A/B tables.

Checklist

• make test and make lint pass locally (docs gates above; no code changed)
• Tests added or updated for the change (none: documentation)
• Documentation extended where it already covers the surface
• Changelog entry added as changes/unreleased/perf-0.11.0-record.added.md (wording updated), not as an edit to CHANGELOG.md
• baselines regenerated and make docs-counts run if a gate count moved (none moved)
• No internal work-item labels in the body, docs, or changelog

Link to Devin session: https://nasa-jpl-demo.devinenterprise.com/sessions/cf22d7a746814d20b2f0eaf11682fef9
Open in Devin Desktop: https://nasa-jpl-demo.devinenterprise.com/desktop/session/cf22d7a746814d20b2f0eaf11682fef9?variant=devin
Requested by: @HuiJun

[devin-ai-integration]

I'll fix CI failures and address comments from users with write access. I'll skip comments containing "(aside)".

• Disable automatic comment, CI, and merge conflict monitoring

[devin-ai-integration]

🔍 Devin Review: 4 flags

Not posted on this PR by your GitHub settings — view them in Devin Review. (Configure)