2598 Commits

This Branch

This Branch

All Branches

Author	SHA1	Message	Date
jschoubben	d871258f8b	Hold that the state error clears once a write succeeds again (issue 1507, review)	2026-10-28 10:08:13 +01:00
jschoubben	e7abe3e3e4	messenger: clear the state problem once a write succeeds again, so the status stops saying a failure that passed (issue 1507)	2026-10-28 10:05:09 +01:00
jschoubben	6cea12a023	messenger: say what waits in the outbox in the history, so an ask held for its burst window does not read as lost (issue 1507)	2026-10-28 10:02:27 +01:00
mesh-tools	3fb7951524	Merge pull request 'claude-code: render a 660 s timeout for the mesh console, so mesh_wait is not cut at a minute (issue 1871)' (#533) from fix/1871-wait-returns-early into main	2026-10-28 07:46:26 +00:00
jschoubben	ae96afcaa8	Render a 660 s timeout for the mesh console and the MCP idle bound, so mesh_wait is not cut at a minute (issue 1871)	2026-10-28 03:52:32 +01:00
mesh-tools	0c63b4c906	Merge pull request 'mesh-delivery-gate: ask the operator's Approve only once the check has a verdict (issue 1733) — wait for its check before approving this one' (#530) from fix/ask-after-check into main	2026-10-27 23:25:26 +00:00
jschoubben	354d62340d	mesh-delivery-gate: the group re-push row holds no verdict at the new heads, so only the missing-run rule keeps it unasked (issue 1733, review round 2)	2026-10-28 00:07:39 +01:00
jschoubben	dea390ac97	mesh-delivery-gate: the Approve's verdict bound counts from the latest head, restarts with its check, and speaks the operator's time (issue 1733, review round 1) ... A group member made ready in an event handler has no ready run when its sibling is looked at; the bound counted from the older sibling alone, so a re-push past 60 minutes asked at once as unanswered. Such a member now counts as ready now. The bound also counts from a check starting again at the same heads, or its verdict being taken away, so a pass that turns back into running never asks at once as unanswered. Times in the wait and the ask are the operator's zone (Europe/Brussels, as the messenger says them), and a placement change whose pass holds no composition says that no node answered its placement.	2026-10-28 00:05:36 +01:00
mesh-tools	00d155da60	Merge pull request 'Take the placement files out again: every node composing its own fails (issue 1126, revert of #528) — approve now' (#529) from revert/1126-seed-4 into main	2026-10-27 22:59:36 +00:00
jschoubben	89c30c44be	mesh-delivery-gate: while the check gates no merge, ask the Approve only once the check has a verdict (issue 1733) ... The operator, 2026-10-27 23:54: "if you don't want me to approve it, don't send the approval request". The gate asked the Approve the moment a pull request opened, before its check answered, and five changes were approved unchecked. Now the ask waits for the checker's verdict at the heads it covers (a group's verdict of all its heads together; a placement change's verdict ends only once its nodes answered, ADR 0559 §5). A pass or a failure is asked, the failure in its words, since check-gates-merge false lets the operator merge it anyway. A check with no verdict 60 minutes after the heads became ready is asked anyway, saying so, and asked again in new words when its verdict comes. Quick fix (only the ask's wording) against thorough (the one place that decides when to ask, ungatedWaits): thorough, which is one place here.	2026-10-27 23:59:09 +01:00
jschoubben	bd65c839a8	Take the placement files out again: every node composing its own fails (novox: nats' ${dir:mesh-state} not declared; ace, g14, shanks: the build index unread at a bus pause), so each returns to the controller's compositions (issue 1126)	2026-10-27 23:56:08 +01:00
mesh-tools	b7a036e186	Merge pull request 'Seed the placement files a third time (issue 1126) — Approve ONLY after the lead says all four nodes answer 'same'' (#528) from feat/1126-seed-3 into main	2026-10-27 22:53:41 +00:00
jschoubben	3abf7cd50b	Seed the placement files a third time, now that every node agent reads its placement only on a successful read and its shadow names every difference (issue 1126, ADR 0549 §4)	2026-10-27 23:52:35 +01:00
mesh-tools	d7e8a8824e	Merge pull request 'Take the placement files out again: since the seed no delivery reaches any node (issue 1126, revert of #525)' (#527) from revert/1126-seed-3 into main	2026-10-27 22:48:46 +00:00
jschoubben	83e9ce5b19	Take the placement files out again: since the seed no delivery reaches any node, because every node agent runs the controller's compositions while the wants ask them to compose their own (issue 1126, revert of #525)	2026-10-27 23:47:52 +01:00
mesh-tools	653cead338	Merge pull request 'Seed the placement files again from what each node runs now (issue 1126, ADR 0549 §4) — Approve only after every node answers 'same'' (#525) from feat/1126-seed-2 into main	2026-10-27 22:02:13 +00:00
jschoubben	d1ca199265	Seed the placement files again from what each node runs now, so placement lives in the catalogue (issue 1126, ADR 0549 §4)	2026-10-27 22:04:46 +01:00
mesh-tools	bf94af0a82	Merge pull request 'gitea: gitea_team_rights reads a forge team's members and units and a user's effective rights, read-only (issue 1563)' (#524) from fix/1563-team-rights into main	2026-10-27 21:00:57 +00:00
mesh-tools	476a59162f	Merge pull request 'mesh-build-go, mesh-build-node: a missing bubblewrap fails the sandbox proofs on a build node instead of skipping them (issue 1309)' (#523) from fix/1309-bwrap-fatal into main	2026-10-27 21:00:53 +00:00
jschoubben	36d5993323	gitea_team_rights: refuse a forge user that does not exist; keep ForgeRepository's doc comment on it (issue 1563)	2026-10-27 21:54:51 +01:00
jschoubben	f8ddcc98f6	Add gitea_team_rights: read a forge team's members, units and a user's effective rights, with the administrator's login, GET only (issue 1563)	2026-10-27 21:50:59 +01:00
jschoubben	9109524502	A missing bubblewrap fails the sandbox proofs on a build node instead of skipping them ... The levels a build module judges hand the tests MESH_BUILD_NODE=1; the tests that run a step in bubblewrap fail on it when bwrap is missing. Issue 1309	2026-10-27 21:36:28 +01:00
mesh-tools	9bd8247915	Merge pull request 'mesh-issues: cover a merged fix from the delivery view's reached head, not only its once-said event (issue 1950)' (#522) from fix/1950-delivered-on-machines into main	2026-10-27 20:32:21 +00:00
jschoubben	116751130f	Cover a merged fix from the delivery view's reached head in the sweep, not only from its once-said event, so a fix named after the saying is delivered (issue 1950)	2026-10-27 21:17:53 +01:00
mesh-tools	c05aa5e6ea	Merge pull request 'Take the placement files out again, so every node returns to the controller's compositions (recovery, issue 1126)' (#521) from revert/1126-seed into main	2026-10-27 14:06:34 +00:00
jschoubben	b95b0a4750	Take the placement files out again: the node agents cannot watch their haves with '>' (refused by their grant), so no node can compose its own; without its placement key each node returns to the controller's compositions (issue 1126)	2026-10-27 09:23:26 +01:00
mesh-tools	04fa2e3286	Merge pull request 'Seed the placement files from what each node runs today, so placement lives in the catalogue (issue 1126, ADR 0549 §4)' (#520) from feat/1126-seed into main	2026-10-27 08:20:33 +00:00
jschoubben	0c61bad102	Seed the placement files from what each node runs today, so placement lives in the catalogue (issue 1126, ADR 0549 §4)	2026-10-27 08:34:10 +01:00
mesh-tools	02597b66d7	Merge pull request 'The issue tracker learns the merge of a fix it never heard, so an old resolved issue no longer reads 'nothing heard from the delivery' (issue 1950)' (#517) from fix/1950-old-fixes into main	2026-10-27 07:21:13 +00:00
mesh-tools	457c60972c	Merge pull request 'The placement check and the delivery gate read the group's mesh-sdk, so replicated seats pass and placement pull messages are read (issue 1126)' (#519) from feat/1126-l2b into main	2026-10-27 07:21:09 +00:00
mesh-tools	5ae337b14d	Merge pull request 'Every build module clears at start what a previous process left running, held by the catalogue's check (issue 1890)' (#518) from fix/1890-step-owner into main	2026-10-27 07:19:39 +00:00
jschoubben	2ac511ce6b	Pin checks/placement and mesh-delivery-gate to mesh-sdk cadd7b1, the group's final head (issue 1126)	2026-10-27 02:44:12 +01:00
jschoubben	e955b53bf0	Pin mesh-delivery-gate to mesh-sdk 062c29c, so it reads a catalogue pull message that carries placement files instead of refusing it (issue 1126; review round 1)	2026-10-27 02:43:13 +01:00
jschoubben	454ac0fec7	Pin checks/placement to mesh-sdk 062c29c, the L2b group's head, so the catalogue's placement check judges with the replicated-seat rule the seed needs (issue 1126)	2026-10-27 02:30:16 +01:00
jschoubben	3c6f2dc274	mesh-issues: a learned merge is kept as of now, closed untimed issues get no fix time, the forge read is declared, the sweep rotates its asks (issue 1950)	2026-10-27 01:30:37 +01:00
jochen	39419b1892	Pin the build modules to mesh-sdk db1007ff, which deletes a gone process's ask key so its redelivery still answers (issue 1890)	2026-10-27 01:30:31 +01:00
jschoubben	b51f03c4ff	mesh-issues: a fix whose pull request has no record learns its merge from the delivery or the forge, so a delivered event can cover it (issue 1950)	2026-10-27 01:24:57 +01:00
jochen	3d42988afa	Hold every build module to ending what a previous process left running (issue 1890) ... A replaced mesh-build-go left a check ask's step running and its start sweep spared it (issue 1890, note 21). mesh-build-go, mesh-build-node and mesh-build-image pin mesh-sdk 981a35b5, whose steps name their owner and whose sweep ends another owner's at once; each runs the sweep through one method that main calls and that its test runs through the conformance kit's left-running case. The failurekit check now refuses a build module whose tests do not run that case, test step or not.	2026-10-27 01:22:13 +01:00
mesh-tools	64530a8dcb	Merge pull request 'Every build module runs the failure contract's conformance kit, and the catalogue refuses one that does not (issue 1901)' (#516) from fix/1901-failure-contract into main	2026-10-26 23:44:16 +00:00
jschoubben	e8a953818c	Merge main into fix/1901-failure-contract: pin mesh-sdk a2759ec (the group's sdk head), the failure kit check's marker lists what it reads (issue 1936), and the registry test's password can no longer match a random workspace name (issue 1901)	2026-10-27 00:39:35 +01:00
mesh-tools	f0f7b6737e	Merge pull request 'Delivery records are true at the source: the gate keeps no state after the merge, and the issue tracker reads delivered from the view (issue 1280)' (#515) from fix/delivery-records-true into main	2026-10-26 21:48:46 +00:00
jschoubben	c1cd8461bc	Merge main (#514, issue 1733) into the delivery-records group	2026-10-26 22:29:36 +01:00
mesh-tools	53bf4bb84c	Merge pull request 'The merge Approve says the check's real state, and is asked again when it passes or fails (issue 1733)' (#514) from fix/1733-ask-says-the-check into main	2026-10-26 21:25:37 +00:00
jschoubben	4677b8a2f5	mesh-delivery-gate: pin mesh-sdk 68d65e6, the group's sdk head merged with sdk main (issue 1280)	2026-10-26 22:21:21 +01:00
jschoubben	cc6e478c4f	Merge remote-tracking branch 'origin/main' into fix/delivery-records-true	2026-10-26 22:21:12 +01:00
mesh-tools	c17ab34a28	Merge pull request 'A reads-outside marker lists what its tests read, so its package's check is reused until one of those files changes (issue 1936)' (#513) from fix/1936-declared-reads into main	2026-10-26 21:18:14 +00:00
jschoubben	099605ec70	mesh-delivery-gate: the Approve's ask says the check's actual state at its heads, never a pass it has not had (issue 1733) ... With check-gates-merge false a head is ready and asked before any verdict exists, yet the ask said "<pr> passed its checks" (a group: "passed its checks"): on 2026-10-26 the operator approved mesh-delivery #79 and a six-repository group on that claim while their checks had not answered. The ask now reads the checker's verdict at exactly the heads it covers (checkAt, beside checkerAt): "passed its check" only for a pass there; otherwise "its check has no verdict yet[: it is running (since HH:MM UTC)]" or "its check failed: <first failure>", and the Approve clause says it merges "now, before its check ends" or "although its check failed". While the check gates no merge, a pending ask whose check then passes or fails (or loses a pass it named) is withdrawn and asked again in new words (the ask's AskCheck against the state now, endStaleAsks, the review-record path's withdrawAsk); a check that starts running is no new word, so the wait stays one message. Whether a merge waits is unchanged. The gate's log lines and stalled lines that claimed a pass for a ready head say the same state. Quick fix: replace "passed its checks" by neutral words in the two summaries. Thorough fix: one reader of the check's state at the ask's heads (checkAt) used by every text that names it, and the ask re-said when that state changes. Chosen: thorough, as it costs one helper and one case in endStaleAsks, and the quick one would still leave a pending ask silent about a failure that came after it. Tests: rows in TestAnApprovedPullRequestIsMergedThroughTheMergeTool (gated pass; ungated no verdict, running, then passed, then failed) and the ungated group's wording; no new test function.	2026-10-26 22:18:08 +01:00
jschoubben	488b7bfc58	checks/health: its marker names health-undeclared for what it counts (issue 1936)	2026-10-26 22:13:14 +01:00
jschoubben	d58249d0d8	checks/health: its marker lists the predefined keys, the modules' templates and health-undeclared it reads (issue 1936) ... ReadTreeManifest reads predefined.json and each templated module's templates (WithTemplatesFrom), and the test reads the root's health-undeclared through Kept: none was listed, so a change to them reused a stale verdict. No other marker's package reads manifests through ReadTreeManifest beyond what it lists. Review finding F1 on ba49a7f0.	2026-10-26 22:13:01 +01:00
jschoubben	d224b74ff1	mesh-build-go: a marked package's digest hashes every file of its own directory, Go files its build tags exclude too (issue 1936) ... go list leaves a file its build tags exclude (an integration test) in IgnoredGoFiles, which the digest did not hash; mesh-provisioner-client's unit test globs *_test.go and reads two of them, so a change there reused a stale unit pass. Review finding F2 on ba49a7f0.	2026-10-26 22:13:01 +01:00