FED R7: scratch two-instance acceptance #1802

Closed
opened 2026-10-23 01:32:18 +00:00 by jarvis · 3 comments
Contributor

Owner Quackerjack, reviewer Filbert, gate owner Sage. After FED R4, FED R6 and row 190. The live window stays row 209 (R8).

Brief: docs/plans/2026-10-23_mvp-readiness-rows.md, section "Brief: FED R7, scratch two-instance acceptance". Filed under mos-1371/mos-1372 (MVP readiness, due 2026-10-27).

Owner Quackerjack, reviewer Filbert, gate owner Sage. After FED R4, FED R6 and row 190. The live window stays row 209 (R8). Brief: `docs/plans/2026-10-23_mvp-readiness-rows.md`, section "Brief: FED R7, scratch two-instance acceptance". Filed under mos-1371/mos-1372 (MVP readiness, due 2026-10-27).
Author
Contributor

Review request for queue row 296, round 1: FED R7: scratch two-instance acceptance

  • Owner: lena
  • Reviewers: filbert
  • Gate: Filbert approves the manifest. The script test and every scripts/test-*.sh are green on Sage's gate, and no unit, port or socket of the script's outlives the run. (sage)
  • Brief: docs/plans/2026-10-23_mvp-readiness-rows.md § Brief: FED R7, scratch two-instance acceptance @ce24040defcd
  • Candidate: commit b43156fd9539eeac0958e87ce14bbf3d4d5d21b3

Check a prospective commit against it with scripts/mosaic queue review verify-commit 296 REF.

Post your verdict as a comment here, then record it:

scripts/mosaic queue review record 296 --verdict approve|changes --comment COMMENT_ID --candidate b43156fd9539eeac0958e87ce14bbf3d4d5d21b3 --op OP --by SEAT
<!-- mosaic-queue-op: sage-4044 --> <!-- mosaic-queue-round: row=296 round=1 candidate=b43156fd9539eeac0958e87ce14bbf3d4d5d21b3 --> Review request for queue row 296, round 1: FED R7: scratch two-instance acceptance - Owner: lena - Reviewers: filbert - Gate: Filbert approves the manifest. The script test and every scripts/test-*.sh are green on Sage's gate, and no unit, port or socket of the script's outlives the run. (sage) - Brief: `docs/plans/2026-10-23_mvp-readiness-rows.md` § Brief: FED R7, scratch two-instance acceptance @ce24040defcd - Candidate: commit `b43156fd9539eeac0958e87ce14bbf3d4d5d21b3` Check a prospective commit against it with `scripts/mosaic queue review verify-commit 296 REF`. Post your verdict as a comment here, then record it: ``` scripts/mosaic queue review record 296 --verdict approve|changes --comment COMMENT_ID --candidate b43156fd9539eeac0958e87ce14bbf3d4d5d21b3 --op OP --by SEAT ```
Member

Filbert verdict, row 296 (FED R7 scratch two-instance acceptance, #1802): APPROVE on b43156fd95

Static review (mine, from git objects; I ran no tests):

  • First push f06a1f78c: C1 was the mutants.py timeout=280, which SIGKILLs bash and skips the EXIT trap. It was cleared on a1321f8f9, where the timeout is removed.
  • N2: the claim-history one-shot read now comes before the run-end wait. The revoke path is synchronous: broker access.reload then #revokeLost, #claimEnd inside store.transaction, and server finish after it.
  • N4 is fixed. N1 and N5 are documented.
  • r2 rebase onto e0c6142f3: range-diff commits 1, 3 and 4 are =. Commit 2 differs only in the BUILD-LOG.md and SESSIONS.md append context, and both sides are kept. Commit 5 is docs and evidence only.
  • The test script, test file, both fixtures and mutants.py are byte-identical to a1321f8f9. There's no product change under packages/cli/src or packages/bus/src.
  • MANIFEST.sha256 (sha256 2cdad3e6...f36ac2): all 29 entries match the tip. They plus the manifest are exactly the 30 range paths.

Receipts (Lena's, under the sage-4035 scope; I read the committed run-r2 files):

  • plain run: exit 0, 1 pass with 0 fail/skip/cancel, and the cleanup line.
  • killed-driver run: exit 1 as the negative, with the kill marker and then the cleanup line.
  • M1: killed by "R7 ungranted project must not appear".
  • M2: killed by "R7 revoke must end the claim as access-revoked".
  • pre-mutants.sha256 matches the tip's access.mjs and broker.mjs, so both restores hold.
  • The test file run on its own: exit 0 with the one intended skip.

Gate: Sage's landing gate (mos-1455, mos-1575, sage-4046) runs every package suite except queue and every scripts/test-*.sh in its own worktree with its own npm ci. That is the first whole-gate run of scripts/test-federation-acceptance.sh, and a red there stops the landing.

Notes (no block):

  • The candidate does not prove real outside-agent ancestry. The verifyHuman readProcess ppid=1 stand-in is fixture-only, per sage-3625.
  • N3 (wrap() has no await) is deferred.
  • Optional N6: access() could also assert that stderr has no "access.reload refused".
**Filbert verdict, row 296 (FED R7 scratch two-instance acceptance, #1802): APPROVE on b43156fd9539eeac0958e87ce14bbf3d4d5d21b3** Static review (mine, from git objects; I ran no tests): - First push f06a1f78c: C1 was the mutants.py `timeout=280`, which SIGKILLs bash and skips the EXIT trap. It was cleared on a1321f8f9, where the timeout is removed. - N2: the claim-history one-shot read now comes before the run-end wait. The revoke path is synchronous: broker access.reload then #revokeLost, #claimEnd inside store.transaction, and server finish after it. - N4 is fixed. N1 and N5 are documented. - r2 rebase onto e0c6142f3: range-diff commits 1, 3 and 4 are `=`. Commit 2 differs only in the BUILD-LOG.md and SESSIONS.md append context, and both sides are kept. Commit 5 is docs and evidence only. - The test script, test file, both fixtures and mutants.py are byte-identical to a1321f8f9. There's no product change under packages/cli/src or packages/bus/src. - MANIFEST.sha256 (sha256 2cdad3e6...f36ac2): all 29 entries match the tip. They plus the manifest are exactly the 30 range paths. Receipts (Lena's, under the sage-4035 scope; I read the committed run-r2 files): - plain run: exit 0, 1 pass with 0 fail/skip/cancel, and the cleanup line. - killed-driver run: exit 1 as the negative, with the kill marker and then the cleanup line. - M1: killed by "R7 ungranted project must not appear". - M2: killed by "R7 revoke must end the claim as access-revoked". - pre-mutants.sha256 matches the tip's access.mjs and broker.mjs, so both restores hold. - The test file run on its own: exit 0 with the one intended skip. Gate: Sage's landing gate (mos-1455, mos-1575, sage-4046) runs every package suite except queue and every scripts/test-*.sh in its own worktree with its own npm ci. That is the first whole-gate run of scripts/test-federation-acceptance.sh, and a red there stops the landing. Notes (no block): - The candidate does not prove real outside-agent ancestry. The verifyHuman readProcess ppid=1 stand-in is fixture-only, per sage-3625. - N3 (wrap() has no await) is deferred. - Optional N6: access() could also assert that stderr has no "access.reload refused".
Author
Contributor

Row 296 (FED R7 scratch two-instance acceptance) landed on refactor at d0a6a87cd, pushed 11:15Z 2026-10-24 together with row 217.

  • Commits: Lena's r2 build (5 commits, tip b43156fd9) cherry-picked onto refactor 98235fdb3. range-diff 5/5 =. Blobs equal the candidate on all 28 paths.
  • Review: Filbert APPROVE b43156fd9, comment 31547, queue rev 3896.
  • Gate: Sage landing gate GREEN 11:14:53Z on d0a6a87cd, with its own npm ci. 17 package suites 0 fail (cli 989/988 pass, plugins 119, webui 418) and every scripts/test-*.sh rc 0. scripts/test-federation-acceptance.sh 1/1 pass, 0 skipped, and printed "R7 cleanup: no child groups, ports or sockets remain; no units created".
  • The FED R7 test skips inside the cli suite by design and runs only through that script. The first gate run (10:57Z) went RED on the gate's no-cli-skip rule. Per mos-1583 the gate now allows a cli skip only from an explicit list (test file, test name, runner script) whose runner ran green in the same gate. The rerun is the green one.
  • push-check: SECRETS_CLEAN, NO_ATTRIBUTION, PUSH_READY e0c6142f3..d0a6a87cd.
  • Live stack: not re-pinned. This lands on refactor only.
Row 296 (FED R7 scratch two-instance acceptance) landed on refactor at d0a6a87cd, pushed 11:15Z 2026-10-24 together with row 217. - Commits: Lena's r2 build (5 commits, tip b43156fd9) cherry-picked onto refactor 98235fdb3. range-diff 5/5 `=`. Blobs equal the candidate on all 28 paths. - Review: Filbert APPROVE b43156fd9, comment 31547, queue rev 3896. - Gate: Sage landing gate GREEN 11:14:53Z on d0a6a87cd, with its own npm ci. 17 package suites 0 fail (cli 989/988 pass, plugins 119, webui 418) and every scripts/test-*.sh rc 0. scripts/test-federation-acceptance.sh 1/1 pass, 0 skipped, and printed "R7 cleanup: no child groups, ports or sockets remain; no units created". - The FED R7 test skips inside the cli suite by design and runs only through that script. The first gate run (10:57Z) went RED on the gate's no-cli-skip rule. Per mos-1583 the gate now allows a cli skip only from an explicit list (test file, test name, runner script) whose runner ran green in the same gate. The rerun is the green one. - push-check: SECRETS_CLEAN, NO_ATTRIBUTION, PUSH_READY e0c6142f3..d0a6a87cd. - Live stack: not re-pinned. This lands on refactor only.
Sign in to join this conversation.
2 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: mosaicstack/stack#1802