[jeremi]

Pull Request

Summary

Record v0.44.0 as published and move the supported documentation selector to its exact source. Open 0.45.0 development across workspace, lockfile, client, editor and generated API version surfaces.

Move the forward-state rehearsal floor to v0.44.0, the published predecessor of the new minor development line. Current-build facade and rstack fixtures derive their versions from the staged package or compiled launcher, preserving version checks across the development bump.

Evidence

• The published tag dereferences to the recorded source, bae18995d0f00eba35d7b96629a56e051ac3d0df. Publication, docs promotion and identifier publication succeeded. Maintained public verification passed for 99 assets, 96 checksum payloads, ten OCI digests, seven registry packages and the macOS Evidence bundle smoke.
• python3 -m unittest release/scripts/test_rehearse_upgrade.py: 86 passed, including v0.44.0 selection and refusal of v0.43.0 before state or download work.
• python3 -m unittest release/scripts/test_registry_release.py release/scripts/test_registry_release_plans.py: 179 total, 178 passed and one platform skip. The plain-binary Scheduling installer fixture is Linux-only; the suite has separate macOS installer coverage.
• Client registry and adjacent packaging unit suites: 119 passed (41 registry, 15 Node sync, 13 wheel assembly, 23 package assembly, ten package smoke, 12 Linux Node helper and five Linux Python helper tests). python3 release/scripts/sync-registry-client-node.py --check passed. The current-source facade staging fixture reads the staged package version; its focused test reproduced the old hardcoded-version failure and passes with the correction. The production version guard and historical boundary tests are unchanged.
• cargo test --locked -p registry-rstack -- --test-threads=1: 83 passed, including 29 bootstrap, dispatch and manager cases. The old fixtures reproduced the compiled-launcher version refusal before correction. A new checksum-valid wrong-version bootstrap test verifies exit 3, the exact error, and absence of a selected generation or public launcher. Formatting and all-target package clippy with warnings denied passed.
• registry-release validate-current, validate-docsets --published-releases ... against the live release list, and cargo metadata --locked --no-deps --format-version 1 passed. The graph has 83 local workspace packages at 0.45.0 and two independently versioned Discovery packages at 0.44.0. External dependency metadata is unchanged.
• Docs npm test: 938 passed. npm run check:docset, npm run check:release-manifests and full npm run check passed.
• git diff --check and redacted Gitleaks of the final diff passed.

Notes

Discovery runtime and operator-tool versions remain independently at 0.44.0. SDK compatibility records advance only the SDK package version; their tested server baseline remains 0.44.0 and their compatible patch line remains 0.44.x. Held Render binding metadata follows the workspace version so local dependency resolution stays valid; its release hold still applies.

This records publication and opens development metadata. Historical docsets, archive locks, release notes and published bytes remain pinned. The floor change adds no runtime schema or migration code.

Local Rust compilation covered the rstack package. Protected CI remains required for broader workspace, generated-exporter, native client and editor compilation.

Verification limit: the initial normal parallel rstack package run hit an intermittent failure in the untouched closed_report_output_is_an_operational_failure test on macOS (expected exit 3, observed 0). Its exact isolated rerun and the final full serial package run passed. The concurrent failure remains unexplained; protected Linux CI must run its normal lane.

DCO

• Every commit includes a Signed-off-by trailer.
• I reviewed the submitted changes and am responsible for the contribution.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-27T22:17:46.518917Z	dc2761b	New commits

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.