Repository navigation
release: 0.68.1 - #3450
Merged
Merged
release: 0.68.1#3450
Conversation
bump-version.sh 0.68.1 (every workspace, facades included; --check green) and the CHANGELOG [0.68.1] section, per docs/specifications/06x-release-schedule.md §4.2. After this merges, rel-068-1-autopilot runs APR-RELEASE-001 §4 T-1..T-4 + close: deep (T-1, local), pre-publish dogfood, tag + release, clean-room.yml dispatched on the tag (T-3, run id recorded), assets by command, preflight, cascade (T-4, automated), install, host + installer receipts, close. Pmat-Ticket: PMAT-1098 Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
§13.11 rung 1 — quorum shadow verdict Shadow mode: this records a verdict and merges nothing. A refusal |
Pmat-Ticket: PMAT-1098 Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
noahgift
enabled auto-merge
September 17, 2026 12:38
Member
Author
|
PMAT-3445 (#3445) — the defect is live on this train. Measured 2026-09-17 ~13:45Z:
When this PR merges, The gate ( |
Merged
guyernest
pushed a commit
to guyernest/aprender
that referenced
this pull request
Sep 29, 2026
…check_milestone_cut.sh at the freeze and before the tag (paiml#3455) * PMAT-3445: plan — read the milestone at the cut, not only at T-5 The 0.68.x autopilot tag step reads no milestone; the only read is the close step, after publish. Plan for a milestone gate, a versioned tag step that calls it, and the spec/contract wiring. Grilled by a width-3 quorum before code lands. Pmat-Ticket: PMAT-3445 Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * PMAT-3445: check_milestone_cut.sh — the milestone being cut must hold 0 open items at the cut v0.68.0 was tagged while milestone 0.68.0 held paiml#3091 (P1, reopened 85 min before the tag) and paiml#3450. The autopilot's tag step read no milestone; the only read was the close step, after publish. This gate exits 0 only when the named milestone has zero open issues AND pull requests, exit 1 naming each open item with its carry/close remedy, exit 2 when it cannot judge (unresolvable or duplicate title, empty milestone, a listed-vs-open_issues disagreement, gh missing/unauthenticated, a failed read). Plan v2 applies the width-3 grill's 3/3 consensus: carry = re-milestone (no comment/event scraping), no release_tag.sh (guard_tree runs only check_*.sh; it would be dark or refused as unwired), every read paginated as JSON lines. Self-test (also the bare run): 18 checks, S1..S16, no network (gh stubbed). Mutations, each on a scratch copy: (a) judge always passes -> RED: S2 S3 S4 S12 S15 S16 (6 of 18) (b) count cross-check deleted -> RED: S6 S7 (2 of 18) (c) --paginate dropped (items) -> RED: S15 (1 of 18) bashrs lint --level error: exit 0. First-green proof, live 2026-09-17: check_milestone_cut.sh 0.67.0 -> 0 (0 open, 50 closed) check_milestone_cut.sh 0.66.0 -> 0 (0 open, 10 closed) check_milestone_cut.sh 0.68.0 -> 1 (paiml#3091 issue [bug,P1], paiml#3450 pr) check_milestone_cut.sh 0.69.0 -> 1 (48 open at the read; milestone API open=48) Pmat-Ticket: PMAT-3445 Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * PMAT-3445: both release specs read the milestone at the cut; contract equation milestone_settled_at_cut 06x-release-schedule.md §4 step 1 reads the milestone at the freeze (before the bump PR opens) and step 4 re-reads it immediately before the tag (after the bump merges) — the binding read, the one that would have seen paiml#3091 reopened 85 min before v0.68.0. Carry = re-milestone plus a `slipped_from:` comment, never an item left open (step 8 closes on open_issues). APR-RELEASE-001 §4 T-0 and T-3 rows carry the same two reads. contracts/release-schedule-06x-v1.yaml: equation milestone_settled_at_cut, RS0-INV-007, FALSIFY-REL-06X-009 (the self-test) and -010 (both specs name the gate), qa_gate check. Verified: pv validate (pinned pv 0.68.0 via verifier_pin.sh) 0 errors; pv lint PASS; FALSIFY-010 exit 0; cargo test -p aprender-core --test readme_contract 15 passed; guard_tree.sh --dry-run --no-cargo prints `run: scripts/check_milestone_cut.sh`; check_guards_are_wired, check_no_timing_in_required, check_assertions_exclude, check_pass_grep_anchored, check_sourced_libs_option_neutral, check_bashrs_gate all exit 0. Pmat-Ticket: PMAT-3445 Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * PMAT-3445: implementation receipt and estimates row Verification table (claimed vs rerun), dispatch ledger with the 3/3 v1 grill and the superseded goal lane, the autopilot-tag rehearsal (STOP, 0 write calls), I-3 attempted=2 running_peak=2 slots=3, findings F-1..F-8. The untracked autopilot tag step is filed separately; it is named in the receipt. Pmat-Ticket: PMAT-3445 Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * PMAT-3445: admit the train's own release epic — it is open at every cut by design Found at pre-merge review (delegate-measured, orchestrator re-measured): each train's release epic sits in its own milestone (06x §5) and is closed at §4 step 8, after publish — paiml#3078 was open in milestone 0.67.0 at the v0.67.0 tag (tagged 07:35Z, epic closed 09:14Z). The binding pre-tag read would have been RED at every cut. The gate now admits exactly one item: an issue labelled `epic` whose title is literally `EPIC: release train <M>` followed by end or whitespace, printed as ADMITTED; two claimants exit 2. A topic epic, a PR, an unlabelled look-alike and another train's epic stay ordinary open items. Self-test 25 checks (S17-S23 added). Mutations on scratch copies: (d) admission removed -> RED S17 S22 S23 (e) admission widened to any `epic` label -> RED S18 S20 S21 (a)(b)(c) still RED (a: 11 checks, b: S6 S7, c: S15) Live: 0.69.0 -> 1, ADMITTED paiml#3080 + 24 other open; 0.70.0 -> 1, ADMITTED paiml#3081 + 28; 0.67.0/0.66.0 -> 0. pv validate 0 errors, pv lint PASS, readme_contract 15 passed, FALSIFY-010 exit 0, check_no_claim_literals 0. 06x §4 step 1 no longer claims step 8 "closes on open_issues" (step 8 does not say so); the refusal is attributed to the train driver. Pmat-Ticket: PMAT-3445 Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * PMAT-3445: quorum verdict — 3/3 PASS at b5a09b0 Merge-rail artifact from quorum-review.sh round 2 (gemini-3.1-pro-high, gemini-3.8-flash-high, gemini-3.7-flash-high; author family claude). Commit touches only the verdict so it speaks for its parent. Round 1 at dfe0852 is superseded: it missed the release-epic defect and two of its lanes read lane 1 first (receipt F-9, F-10). Pmat-Ticket: PMAT-3445 Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Release bump for 0.68.1 (06x release schedule §4.2):
bump-version.sh 0.68.1across every workspace, and the CHANGELOG section below.When this merges,
rel-068-1-autopilot/autopilot.shruns the rest of the train, fail-closed (APR-RELEASE-001 §4): T-1 deep run, pre-publish dogfood, tag + release,clean-room.ymldispatched on the tag with the run id recorded (T-3), all release assets checked withscripts/check_release_assets.sh, publish preflight, the crates.io cascade (T-4, automated per operator 2026-09-13),install.sh --versionreceipts on intel and gx10 plus the CUDA asset receipts on gx10 and yoga, and the epic and milestone close.The crates.io release of the 0.68 line.
v0.68.0is a GitHub-only release (tag and binaries stand;install.shverified on x86_64 and aarch64) — its clean-room on the tag passed A0–B1 for the first time ever on a tag and stopped at B2: lib tests inaprender-coreandaprender-test-showcasenamed path-only dev-dependencies thatcargo publishdeletes (#3425). 0.68.1 carries that fix — the tests moved out ofsrc/to targets the published crates do not carry, all 72 lib-test binaries build in published form — plus the two infra gate repairs that made the chain provable (infra#650 tag ref materialized, infra#652 A1[patch.crates-io]overlay at the tag). Everything else is 0.68.0: Qwen 3.5 on the CPU, the one-line installer, the release-train gates.Fixed
Changed
🤖 Generated with Claude Code
no-close: #650 652 3415 3417 3419 3425 3426 — referenced by the 0.68.1 release notes; the bump PR closes nothing.