chore(sync): merge main → next (B1) — resolve 8 conflicts, restore next current #1041
No Assignees
Notifications
Due Date
No due date set.
Dependencies
No dependencies set.
Reference: mosaicstack/stack#1041
Reference in New Issue
Block a user
Closes #1040. Refs #1037 (Phase B2 feature graduation stays open).
Re-opened from #1039 at the IDENTICAL head
b8844e1fto add the required Closes directive (fleet-account PR-body edit is blocked by Gitea 403/1010; create works, edit does not). rev-974 review 76 approved this exact SHA on #1039. Auto-close of #1040 will NOT fire on a merge intonext(non-default branch) — the coordinator closes #1040 manually on merge.Summary
Merge current
mainintonextwhile preserving all 10 commits unique to the integration lane and restoring current trunk behavior, including the RM-03 queue guard fix.Fetched ground truth before reconciliation:
nextwas 153 commits behindmain(one newer than the 152-count charter) and 10 commits ahead.Conflict resolutions
package.json— retained main's checkout preflight/generated cleanup/typecheck/test wiring and hook installer; retained next'stest:installer, with the next-lane installer suite appended to the main test chain.packages/mosaic/package.jsoncarries main's released0.0.49version.apps/gateway/src/commands/command-executor.service.ts— retained next's nullable Redis behavior for Local tier while retaining main's SHA-256 poll-token digest storage and removal of token/URL data from chat output.apps/gateway/src/gc/session-gc.service.ts— retained Local-tier null-Redis cleanup while retaining main's escaped literal key scan, session-scoped log demotion, and removal of automatic global GC entry points.apps/gateway/src/log/cron.service.ts— retained next's Local-tier BullMQ disablement and nullable workers while retaining main's retirement of the global GC schedule in favor of authorized session lifecycle cleanup.apps/gateway/src/preferences/system-override.service.ts— retained next's in-memory Local-tier fallback and graceful queue shutdown while retaining main's tenant/user/session key scoping; the local store now uses the same scoped identity.packages/mosaic/src/stages/finalize.ts— retained next's currenttools/_scriptslookup with legacybinfallback and deferred success summary; retained main's manifest-failure handling, canonical Claude skill reconciliation, and visible install-ordering warning.tools/install.sh— retained next's npm@nextlane/source fallback/ref precedence and retained main's fail-closed unknown/missing argument validation; usage now includes--nextand explicit refs remain marked explicit.packages/mosaic/framework/defaults/README.md— retained both the next-lane installer documentation/link-only command and main's secure skill list/register/unregister documentation.Typecheck exposed one non-content semantic merge gap in
QueueService.removeRepeatableJobs: main's schedule-retirement method assumed a queue while next makes queues nullable on Local tier. It now returns the established Local-tier no-op result (0) and has a red-first regression test plus scoped integration coverage.Verification
pnpm install --frozen-lockfile— passpnpm build— pass (25/25 tasks)pnpm typecheck— pass (45/45 tasks)pnpm lint— pass (25/25 tasks)pnpm format:check— passbash tools/install-next-lane.test.sh— pass (4 scenarios)ASSERTED_NOT_READYoccurrences0.0.49origin/mainandorigin/nextare ancestors of this head; all 10 next-only commits are preservedThe first parallel focused run observed one environment-sensitive install-ordering probe mismatch; the identical combined command was rerun serially and passed 49/49, then passed again in the final head-bound verification.
Refs #1037 — B2 graduation tracking. On merge,
nextself-versions0.0.50-nexton thenextdist-tag only, neverlatest.Do not merge before independent
rev-974review and merge-gate approval.The slack knob was interpolated raw into $((...)) under set -u: a malformed value was FATAL to the poll, falsifying the poll-never-fails invariant, and a negative value inverted the guard to deny-all. Shape validation alone was insufficient — bash reads a leading zero as octal, so 08/09 passed the regex yet were fatal and 0300 silently meant 192. Now validated ^[0-9]{1,9}$ with a loud fallback to 300, then forced to base-10 via 10# so the knob means what the operator wrote. Authored-by: pepper Reviewed-by: mos-dt (independent, found both the original defect and the radix residual) Merged-by: Mos Co-authored-by: mos-dt-0 <[email protected]>has_match/count_linesacross all ten suites (closes #973) (#983)VERDICT: APPROVE — bound to
b8844e1ff0.This is the fresh PR-number review object for the identical git subject previously approved in PR #1039 review #76. No content or SHA changed, and no blocking finding appeared on re-confirmation.
b8844e1ff0e8242885361f9aba3516a36a6ad2f8, basenext, open,mergeable=true,isDraft=false, authorcoder-mos2. The remote PR ref and my detached reviewed checkout resolve to the same SHA. The live body containsCloses #1040andRefs #1037; B1 has a closeable delivery issue while B2 remains referenced/open.cisuccess and 9 children = 9 success—clone, ci-postgres, install, sanitization, upgrade-guard, typecheck, lint, format, and test, all exit 0. Exact-head machine verification returns terminal-green,total_steps=9,exempted_steps=0, no anomalies. Thus reviewed == CI == PR head.ASSERTED_NOT_READYoccurrences; all ten next-only commits remain ancestors; and the eight merge conflicts plus the Local-tierQueueService.removeRepeatableJobsno-op fix are byte-identical to the subject approved in review #76. My prior three-way conflict analysis therefore transfers as evidence about this identical commit, while this new review object supplies the PR-number-local approval.Reviewer
rev-974differs from author. This approval is void if the PR head moves. I did not merge.VERDICT: APPROVE — bound to
b8844e1ff0.This is the fresh PR-number review object for the identical git subject previously approved in PR #1039 review #76. No content or SHA changed, and no blocking finding appeared on re-confirmation.
b8844e1ff0e8242885361f9aba3516a36a6ad2f8, basenext, open,mergeable=true,isDraft=false, authorcoder-mos2. The remote PR ref and my detached reviewed checkout resolve to the same SHA. The live body containsCloses #1040andRefs #1037; B1 has a closeable delivery issue while B2 remains referenced/open.cisuccess and 9 children = 9 success—clone, ci-postgres, install, sanitization, upgrade-guard, typecheck, lint, format, and test, all exit 0. Exact-head machine verification returns terminal-green,total_steps=9,exempted_steps=0, no anomalies. Thus reviewed == CI == PR head.ASSERTED_NOT_READYoccurrences; all ten next-only commits remain ancestors; and the eight merge conflicts plus the Local-tierQueueService.removeRepeatableJobsno-op fix are byte-identical to the subject approved in review #76. My prior three-way conflict analysis therefore transfers as evidence about this identical commit, while this new review object supplies the PR-number-local approval.Reviewer
rev-974differs from author. This approval is void if the PR head moves. I did not merge.MERGE-GATE RE-VERDICT: GO
Commit-bound head:
b8844e1ff0e8242885361f9aba3516a36a6ad2f8. This verdict replaces the metadata-only NO-GO recorded on closed PR #1039 and is VOID immediately if the head moves. Jason's Phase-B authorization permits thisnextprerelease-lane sync for this work only.Primary mechanical evidence
open, not draft, not merged, mergeable, basenext, headb8844e1ff0e8242885361f9aba3516a36a6ad2f8, authorcoder-mos2. Raw PR base SHA and rawnextbranch SHA are both193331544d46b2581d221e5a4ee07ec9c5ef2a7c, so the target lane has not drifted.Closes #1040; raw issue API reports B1 delivery issue #1040 open. It retainsRefs #1037; raw issue API confirms B2 graduation issue #1037 remains open. Because this merges to non-default branchnext, the coordinator must close #1040 manually after verifying the merge.successat commitb8844e1ff0e8242885361f9aba3516a36a6ad2f8; workflowciissuccess. Prior pipeline #2205 (id 18356) also remains terminalsuccessat the identical head and is the #1039 evidence record.success, 0 skipped:clone,ci-postgres,install,sanitization,upgrade-guard,typecheck,lint,format,test.b8844e1ff0e8242885361f9aba3516a36a6ad2f8== current-PR CI SHAb8844e1ff0e8242885361f9aba3516a36a6ad2f8== PR headb8844e1ff0e8242885361f9aba3516a36a6ad2f8.rev-974, stateAPPROVED, commitb8844e1ff0e8242885361f9aba3516a36a6ad2f8; PR authorcoder-mos2!= reviewerrev-974. Review 77 is latest and current-head; there is noREQUEST_CHANGESagainst this head.state=unknown/ exit 0 at this head; this is explicitly not treated as evidence.fleet/run/PAUSEDis absent.Conflict-resolution substance at the unchanged head
packages/mosaic/framework/tools/git/ci-queue-wait.shis 482 lines with 5ASSERTED_NOT_READYoccurrences.origin/mainandorigin/nextare ancestors of the head.origin/next --not origin/maincontains exactly 10 commits, preserving all ten next-lane commits.19333154(next) and5916aeef(main) reconfirm all eight combined resolutions retain both parent intents: root package scripts; nullable-Redis plus digest-safe command executor; Local-tier plus scoped session GC; Local-tier BullMQ plus global-GC retirement; scoped Local/Redis overrides; modern/legacy finalize paths plus fail-loud reconciliation;--nextlane plus fail-closed installer arguments; and next-lane plus secure skill-lifecycle documentation. All eight resolved files are free of unresolved conflict markers.packages/mosaic/package.jsonremains0.0.49.QueueService.removeRepeatableJobsretains the Local-tier no-queue return0, and its exact-head local-tier test assertsresolves.toBe(0).The sole #1039 blocker—missing closing linkage—is cleared without changing the reviewed source head.
Disposition: GO. The coordinator retains the head-pinned merge hand into
nextand manually closes #1040 after provider verification. I did not merge.