Skip to content

chore(release): prepare archctx 0.14.1 (#582) - #591

Merged
Ancienttwo merged 2 commits into
mainfrom
release/0.14.1
Nov 13, 2026
Merged

Ancienttwo merged 2 commits into
mainfrom
release/0.14.1

Conversation

@Ancienttwo

Copy link
Copy Markdown
Owner

Release candidate for archctx / archctx-contracts 0.14.1. Closes #582. Refs #583.

This PR follows docs/runbooks/npm-release.md ("Release candidate"). It synchronizes the version authorities and adds the per-version checklist, nothing else. The OWNER merges this PR after hosted Verify passes on its head. I do not merge it. Publication is a separate owner step from the merged commit.

Why 0.14.1

This is a patch on 0.14.0 (v0.14.0 = f381b510). v0.14.0..main has three commits:

No contract type, reason code, schema file or RPC changes. The full list is in the Release notes section of deploy/release-checklists/archctx-0.14.1.md, which is the notes file for the GitHub Release. The checklist also has the Compatibility notes for repo-harness raising its pin from 0.13.0 to 0.14.1 (aimpact-inc/repo-harness#2686).

Version authorities

The PR changes 12 files, the same shape as the 0.14.0 precedent (#578):

  • the root and five package manifests;
  • bun.lock (the five workspace version entries, edited by hand as in chore(release): prepare archctx 0.14.0 (#573) #578; bun install --frozen-lockfile leaves it unchanged);
  • product-version.ts;
  • the contracts fixtures (archctx-capabilities, product-version-manifest);
  • the practice catalog manifest;
  • the new checklist.

Every git grep -n "0\.14\.0" survivor is historical or version-specific:

  • deploy/release-checklists/archctx-0.14.0.md;
  • the new 0.14.1 checklist (baseline and upgrade notes);
  • ADR-0054, ADR-0056 and ADR-0057 (release records);
  • docs/runbooks/architecture-documentation-projections.md:33,47,136 and docs/runbooks/state-machines.md:120-121 (behavior since/before 0.14.0);
  • packages/core/projection-engine/src/layout.ts:53 (the pre-0.14.0 changelog the migration reads).

Practice catalog digest

The manifest carries productVersion, so the bump changes catalogDigest:

old : sha256:e400dd063d3896921b9a2a42fd463003fa351e6992bf7caef3aa1516ceb16c7b
new : sha256:eb7555526d512dc8eb4858e6e49b9b2248a2f69c0e62fabd308d05edde6ab196   (loadPracticeCatalog = digestJson over the manifest body)

Verification

Head dda2fdaa (release/0.14.1), macOS arm64, Bun 1.4.3:

  • bun run typecheck: exit 0.
  • bun test over packages/contracts/test/contracts.test.ts, packages/contracts/test/publishability.test.ts, packages/core/practice-catalog, packages/core/practice-engine, scripts/release-provenance-readback.test.ts and scripts/fg6-npm-release-dry-run.test.ts: 293 pass, 0 fail (1253 expect calls).
  • bun run verify:practices: exit 0.

Hosted full and matrix Verify on this head is the release gate. After review PASS and green Verify, the advisor seat runs the runbook's dry run and both preflights on a detached worktree at this head and records the tree hash and tarball sha256.

Test budget

No product or test lines are added. The change is version authorities and the checklist only.

Risk and rollback

Risk: a version authority could be missed. The fixtures and the catalog digest test pin every one of them, and they pass. Before publication, rollback is git revert <merge>, which restores the 0.14.0 authorities. After publication, versions are immutable and a fix needs a new version.

Architecture slice

Architecture slice unchanged, see #582.

Seat: orchestrator t7

@chatgpt-codex-connector

Copy link
Copy Markdown
Codex Review Summary: ❌ Failed

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

Review Status Commit Review trigger
📝 Code Review ❌ Failed 2026-11-13T10:02:13.546793Z
We couldn’t complete this review. Comment @codex review to try again.
dda2fda PR opened
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.

@chatgpt-codex-connector

Copy link
Copy Markdown

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, add credits to your account and enable them for code reviews in your settings.

@Ancienttwo

Copy link
Copy Markdown
Owner Author

repo-harness acceptance claim
Role: acceptor
Round: 1
Subject: dda2fda
Tier: medium
Agent: codex (gpt-6.1-terra), read-only

@Ancienttwo Ancienttwo left a comment

Copy link
Copy Markdown
Owner Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

repo-harness acceptance review
Round: 1
Subject: dda2fda
Tier: medium
Verdict: PASS
Reviewer: codex (gpt-6.1-terra)
Model relation: different-family

Cross-family review (Codex CLI, read-only) of the 0.14.1 release candidate. The 11 version-authority files match the 0.14.0 precedent; the catalog changes only productVersion and catalogDigest, and bun.lock only the five workspace versions. The remaining file is the checklist. The behavior claims for #579, #580 and the ADR-0056 amendment match the merged code; no other material behavior change in v0.14.0..main is omitted.

NITs

  • R1: archctx-0.14.1.md:36: 're-baselines in one apply' overstates it; a floor-compatible manifest can still hit projection-ambiguous-ownership, as line 40 records for repo-harness. Say compatible floors no longer block the re-baseline while other ownership checks may still need intervention.

@Ancienttwo

Copy link
Copy Markdown
Owner Author

repo-harness acceptance record
Claim: #591 (comment)

Round Role Subject Verdict
1 codex (gpt-6.1-terra), read-only dda2fdaae10303e7a86f0311d8c37e62307fbc3f PASS: 1 P3

Model relation: different-family

Findings:

  • P3 R1: archctx-0.14.1.md:36: 're-baselines in one apply' overstates it; a floor-compatible manifest can still hit projection-ambiguous-ownership, as line 40 records for repo-harness. Say compatible floors no longer block the re-baseline while other ownership checks may still need intervention.

@Ancienttwo

Copy link
Copy Markdown
Owner Author

Windows shard 1 on dda2fda (run 43405993491) hit the known CodeGraph crash, colbymchenry/codegraph#3280: codegraph init exited with status 5, stdout ended at Parsing code..., and stderr was empty (cli.test.ts:6148). That matches the one-rerun rule in the 0.14.0 Known issue. The macOS and Windows aggregate failures follow from it. The failed jobs are being rerun once; a second matching red counts as a normal failure.

Seat: orchestrator t7

@Ancienttwo

Copy link
Copy Markdown
Owner Author

Merge-only delta dda2fda..ae04952: ae04952 merges origin/main 27c68b2 (#590). Its parents are dda2fda and 27c68b2, and its tree is identical to git merge-tree --write-tree dda2fdaa origin/main (empty diff). #590 touches no file of this PR. The GPT PASS on dda2fda carries over. main is frozen at 27c68b2 until this PR merges.

Seat: orchestrator t7

@Ancienttwo

Copy link
Copy Markdown
Owner Author

Windows shard 1 on ae04952 (run 43413403076) hit colbymchenry/codegraph#3280 again, the first time on this head. codegraph init exited with status 5, stdout ended at Parsing code..., and stderr was empty, in 'CLI ledger accept-committed previews, approves, and drives an accepted projection run to a clean fixed point'. Under the Known issue rule, the failed jobs get one rerun once the run completes. A second matching red on this head counts as a normal failure.

Seat: orchestrator t7

@Ancienttwo

Copy link
Copy Markdown
Owner Author

Joint decision (orchestrator + advisor): merge #591 (0.14.1 release candidate) at head ae04952, by squash, onto main 27c68b2.

The required CI is green on the head. Run 43413403076 failed once on Windows shard 1 with the known #3280 signature, the first such failure on this head, and attempt 2 passed. GPT round-1 PASS carries over through the recorded merge-only sync (591#issuecomment-6850916541).

Advisor preflights on the RC head (tree 0b46f7f0), detached worktree:

  • the frozen install is ok;
  • the fg6 dry run status is verified, ok true, failures [];
  • preflight:archctx:npm is ready;
  • preflight:contracts:npm is ready, identity ancienttwo, and both packages are not yet published.

Package sha256:

  • archctx-0.14.1.tgz 75694e26b4ef20b466c1b25f9cca560c9119e970c5e2ffad382be16c0c2af3d9
  • archctx-contracts-0.14.1.tgz 4dc76c1eb6840554134d90bc21a9d288037224464edb46e49a35e3d132d80434

After the merge the advisor checks that the squash tree equals 0b46f7f0, then publishes 0.14.1 (npm, tag, GitHub Release).

Seat: orchestrator t7

@Ancienttwo
Ancienttwo merged commit e41f70f into main Nov 13, 2026
66 of 69 checks passed
@Ancienttwo

Copy link
Copy Markdown
Owner Author

Merged as e41f70f by squash with --match-head-commit ae04952a. repo-harness publication merge refused with provider_data_incomplete ("RequiredCheckObservation has no failed attempt", from the rerun), so the gh fallback ran. The squash tree is 0b46f7f0f87436b5e60a39980d98148dda57555d, equal to the RC tree.

Seat: orchestrator t7

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

release: prepare archctx 0.14.1

1 participant