[devin-ai-integration]

What and why

#1740 landed the 0.12.0 performance record with the release candidate measured at cc1d477ba and the export rows re-probed at 5e0322e0c. develop kept moving while it was reviewed, and several of the merges touch measured paths, so the record's "The release candidate" section is amended to account for every merge up to develop e022cbca0:

• fix(export): read one relationship id per feature-chain clause, as the writer materializes them #1774 (fix(export): read one relationship id per feature-chain clause, d01e4f49c) changes internal/translate/export/element_ids.go: the export rows were re-probed a second time, six rounds interleaved with both 0.11.0 columns — WriteAPIJSON and ConvertAPIJSON allocate the same bytes and the same number of times as at 5e0322e0c on every row, and the whole-binary element-ID probe was repeated against the cc1d477ba binary.
• feat(engine): run state machines and actions on the clock (advance), with a clocked hero scene #1777 (feat(engine): run state machines…, 1e34d9d95) changes internal/exec/runtime/context.go and the engine/gRPC frontends: six interleaved rounds of the runtime and gRPC packages, repl LoadModel and the tests/perf rows, cc1d477ba against 1e34d9d95, steal logged per round — bytes and allocations equal on every exact row, InstantiateWarmModel inside its own spread, sec/op ~, the tests/perf peak resident set within spread.
• fix(semantics): Usage::mayTimeVary derives over the effective isComposite — a performed action or exhibited state varies in time #1784 (fix(semantics): Usage::mayTimeVary derives over the effective isComposite, 9a4ffcd83) is on the semantics path every load runs: six interleaved rounds at 9a4ffcd83 against cc1d477ba show it moves the load rows — repl LoadModel/1000 +1.5 % bytes and +4.8 % allocations per element loaded, ImportValues/nested +11–13 % allocations, the runtime and instantiation rows unchanged, sec/op ~. The record names fix(semantics): Usage::mayTimeVary derives over the effective isComposite — a performed action or exhibited state varies in time #1784 as the cause (it takes back about a third of what perf(check): sweep a document's expressions for typed by only while one is unreported #1744 recovered on the load rows) and does not fix it.
• fix(export): the RDF export states a usage's isVariable and mayTimeVary as the metamodel derives them; the element form states the pair both ways #1783 (fix(export): the RDF export states a usage's isVariable and mayTimeVary as the metamodel does, e022cbca0) changes five files under internal/translate/export: the export rows were re-probed a third time, six rounds at e022cbca0 against d01e4f49c — every row grows with its output (WriteAPIJSON output +0.4–2.2 %, bytes +0.06–0.57 %, allocations +0.3–0.9 %; ConvertAPIJSON/32 and /128 +0.7–0.8 % bytes, +1.2 % allocations, all p=0.002), sec/op ~, the whole-binary element-ID query, Turtle export and -validate within spread.
• The client-parity merges, chore(deps): bump the maven-minor-and-patch group across 3 directories with 1 update #1748/chore(deps): bump softprops/action-gh-release from 2.3.3 to 2.6.2 in the actions-minor-and-patch group #1750 (poms, a workflow), fix(ci): check protobuf compatibility against the pull request's merge base #1781 and test(client): narrow a refusal lambda and merge the model imports #1786 are shown runtime-neutral by git diff --stat … -- '*.go' go.mod go.sum.

The candidate is named as develop e022cbca0, with the SHA each row set was measured at (cc1d477ba, re-checked at 1e34d9d95 and 9a4ffcd83, for the runtime/load/REPL/LSP/whole-binary rows; e022cbca0 for the export rows), and the "What this record does not measure" list moves its cut accordingly. The changelog fragment of #1740 gains the words "with every merge up to the release head accounted for".

How it was verified

python3 scripts/check-doc-links.py reports only the pre-existing third-party-notices.md link; python3 scripts/changelog.py check passes; every table in the record has consistent columns. The raw per-round outputs of every probe (steal logged per round, the whole-binary sequences, the tests/perf RSS runs) are kept outside the repository and attached to the gate report, as the rest of the record's are.

Checklist

• make test and make lint pass locally — documentation only; no Go source changes
• Tests added or updated for the change — not applicable
• Documentation extended where it already covers the surface (see CONTRIBUTING.md)
• Changelog entry added as changes/unreleased/<slug>.<section>.md, not as an edit to CHANGELOG.md — the existing perf-0.12.0-record.added.md fragment is amended
• baselines regenerated and make docs-counts run if a gate count moved — no gate count moved
• No internal work-item labels (waves, slices, F4, K5) in the body, docs, or changelog

Link to Devin session: https://nasa-jpl-demo.devinenterprise.com/sessions/15408827f8104c7fbbdbd1b2c852ed2c
Open in Devin Desktop: https://nasa-jpl-demo.devinenterprise.com/desktop/session/15408827f8104c7fbbdbd1b2c852ed2c?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: 1 flag

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