diff --git a/core/decisions/DDD-dec-123.yaml b/core/decisions/DDD-dec-123.yaml new file mode 100644 index 0000000..3ecd3ab --- /dev/null +++ b/core/decisions/DDD-dec-123.yaml @@ -0,0 +1,79 @@ +format: 1 +id: DDD-dec-123 +kind: decision +statement: | + For one landing only, the trunk ruleset's required `agent review` check is lifted, so that #83's fix can reach main; it is restored straight after. Nothing else in trunk or in any other ruleset changes, and nothing in DDD-dec-108 or DDD-dec-109 is superseded. The waiver is not a precedent: a later broken main is a new decision. +principal: Emil +made: | + 2026-11-18, G1 of the canon-lock-file session (issue #83), given by Emil in the session and recorded verbatim at meta/sessions/2026-11-18-canon-lock-file/rulings.md (received 13:17 UTC). Times Europe/Berlin, all Emil's acts as repository admin: the requirement was lifted at 15:20; restored at 15:28 under the wrong check name; corrected at 15:39 (rulings.md, 2026-11-19). Emil's ruling on this record's text (G2, received 13:37 UTC, in the same file) is the filing act +resolution: | + ## The ruling + + > Q-1. I will lift the "agent review" requirement on the trunk ruleset for this one landing and + > restore it straight after. You do not touch the ruleset. + + ## What was waived + + The `required_status_checks` entry `agent review` (integration 5594102, the gates App) in the + live trunk ruleset, for the landing of head 1bf7e56 (1bf7e56af31cd19e004def678c163ccf224c134f) + to main and for nothing else. The trunk ruleset's other rules held throughout: `pipeline + (devcontainer)` required, deletion and non-fast-forward refused. Signed Commits was not + touched. + + .github/repo-standard.yaml was not changed and required `agent review` throughout. Only the + live trunk rule departed from it, from 15:20 to 15:39 Europe/Berlin: + + - 15:20: `agent review` lifted. + - 15:25: 1bf7e56 landed on main. + - 15:28: restored as "agent review (judge)", bound to the gates App. That is the workflow + job's name, which the gates App never posts, so no pull request could have satisfied the + rule. + - 15:39: corrected to `agent review` from the gates App (integration 5594102), which matches + .github/repo-standard.yaml. + + Nothing landed on main between 15:25 and 15:39 other than 1bf7e56. This was checked against + main's history before it was written here. main moved once since 84aa466, a fast-forward to + 1bf7e56, and was still there on 2026-11-19. The only push-triggered run on main in that time + is validate run 392, on 1bf7e56, at 15:25. + + ## Why no other path existed + + - The `agent review` verdict is a check run posted by the gates App ("mom-gates", integration + 5594102), and the ruleset requires that name from that App (DDD-dec-109). No job a pull + request can change may post it, by design. + - The agent review job checks out main by name and builds main's canon before it judges + (.github/workflows/agent-review.yml, "Checkout main"). Main's canon did not build on the + runner's SDK: NU1004, the lock file at Microsoft.NET.ILLink.Tasks 10.0.14, the runner's SDK + 10.0.404 adding 10.0.15 (meta/sessions/2026-11-18-canon-lock-file/findings.md, D-1 to D-4). + - So agent review failed on every pull request whatever it held, the fix's own included, and + trunk has no bypass actor, the admin role included (DDD-dec-108). The fix could reach main + only if the requirement was suspended. + + ## When this record was filed + + After the waiver was used, because it needs the landed head and the two times. Emil's ruling + to waive was recorded first, in this session's rulings.md, which landed with #85 (at + 1bf7e56). His ruling on this record's text is recorded in the same file on this record's + branch. + + ## What showed the fix before it landed + + Validate on the head, with locked restore (CI passed into the devcontainer); the agent review + job's build on a clean install of SDK 10.0.404 with CI set; the same on a GitHub runner by a + throwaway workflow, removed before the head (findings.md, deliverable 3). The first agent review + run after the landing is the final proof: Agent review run 89 (44325821613), on #86 at e28fc52, + 15:34 to 15:35 Europe/Berlin. It built main's canon at 1bf7e56 on SDK 10.0.404 without NU1004, judged the + pull request and posted its verdict. +basedOn: + - DDD-dec-108 + - DDD-dec-109 +notes: | + Successor, recorded by Emil's ruling and not acted on: a required check that builds main's + canon, with no bypass, means a broken main can only be repaired by suspending the rule. This + record is the first time that happened. + + Agent review runs 89 and 90 (#86, 15:34 and 15:38 Europe/Berlin) ran while the live rule + required "agent review (judge)". The rule's name does not change what the workflow runs. Both + built main's canon at 1bf7e56 and posted an `agent review` check from the gates App. Neither + could have satisfied the rule then in force, and both verdicts were `failure` for want of + approval. diff --git a/meta/sessions/2026-11-18-canon-lock-file/predictions.md b/meta/sessions/2026-11-18-canon-lock-file/predictions.md index d6c343d..55575db 100644 --- a/meta/sessions/2026-11-18-canon-lock-file/predictions.md +++ b/meta/sessions/2026-11-18-canon-lock-file/predictions.md @@ -101,3 +101,29 @@ Prediction: on push, the throwaway workflow's setup-dotnet step logs an install Result: **held** for the throwaway run (44313244012: `--version 10.0.404`, `CI=true`, build succeeded, 0 warnings). Validate on `3d0ac65`: run 389, success, with `CI=true` passed in (findings.md, G2). + +## P-8. Filing DDD-dec-123 (before the filing commit) + +Prediction: `core/decisions/DDD-dec-123.yaml`, from the accepted draft with Emil's two W-2 +additions and the head and times filled, passes `validate-claims.py core/decisions/ --decisions` +and the whole validate pipeline (devcontainer, `CI=true`), with no other file under `core/` +changed. On the pull request, `pipeline (devcontainer)` passes, and `agent review` (from `main` +at `1bf7e56`, the pinned SDK) builds canon without NU1004 and posts its verdict. Without Emil's +`approve ` the verdict is not green: the job runs, and fails or stays pending for the +approval, not for the build. + +Result: **held.** `validate-claims.py --decisions`: 97 decisions valid. Local validate on +`e28fc52` (`CI=true`): every gate ok. Agent review run 89 (44325821613) on #86: job success; +canon built from `main` at `1bf7e56` with no NU1004; "not approved: #86 needs a comment 'approve +e28fc52…'"; "posted 'agent review' = failure". The verdict waits for Emil's approval, not for the +build. It was the first agent review run after the landing; #84 had not re-run. + +## P-9. DDD-dec-123 amended for the three times (before the amending commit) + +Prediction: the decision with the three Europe/Berlin times (15:20, 15:28 with the wrong name, +15:39 corrected), the departure window 15:20 to 15:39, and "nothing landed between 15:25 and +15:39 other than 1bf7e56" passes `validate-claims.py --decisions` and the whole validate +pipeline. On #86's new head, agent review builds main's canon at `1bf7e56` and posts `failure` +for want of approval; validate passes. + +Result: (scored below) diff --git a/meta/sessions/2026-11-18-canon-lock-file/rulings.md b/meta/sessions/2026-11-18-canon-lock-file/rulings.md index 7d61e0b..1e09bb6 100644 --- a/meta/sessions/2026-11-18-canon-lock-file/rulings.md +++ b/meta/sessions/2026-11-18-canon-lock-file/rulings.md @@ -51,3 +51,82 @@ Claude's notes, where any, follow a ruling and are marked as Claude's. > > Stop at G2 with a pull request against main, and name the head to ratify. Out of scope stays > as in the issue. + +## G2 — received 2026-11-18 13:37 UTC (14:37 Europe/Berlin) + +> My rulings at G2 on #83. Record this message in rulings.md, in my words, with the time you +> receive it. +> +> W-1. The waiver takes DDD-dec-123. Check every branch of both repositories once more when you +> file it. +> +> W-2. The draft's text is accepted, with two additions: +> - say that the record is filed after the waiver was used, because it needs the landed head and +> the two times, and that my ruling was recorded first in this session's rulings.md, which +> lands with #85; +> - say that .github/repo-standard.yaml was not changed and still requires "agent review" +> throughout: only the live trunk rule departed from it, between the two times. +> +> I will now lift "agent review" on trunk, land 1bf7e56, and restore it with its binding to the +> gates app (integration 5594102). Change nothing and push nothing until I give you the landed +> head and the two times. +> +> When I do: +> 1. Branch from the new main. File core/decisions/DDD-dec-123.yaml with the fields filled, and +> move nothing else. +> 2. Open a pull request against main with "Refs #83" on every commit. It is the first change +> through the restored gate, so report both required checks on its head. +> 3. Fill the "first agent review run after the landing" field with whichever run completes +> first, this pull request's or #84's. If that run fails for any reason other than the +> decision's own content, report it and stop. +> +> Stop there. Merging is mine. + +Claude's note: received while `1bf7e56` awaited landing. Under "change nothing and push +nothing", this was held outside the repository and is recorded here, on the filing branch +(`claude/canon-lock-file-waiver-2026-11-18`, from `main` at `1bf7e56`), not with #85. + +## Landing facts — given by Emil, 2026-11-18 + +> restored - lifted 14:20, restored 14:29 UTC + +Landed head: `1bf7e56` (`1bf7e56af31cd19e004def678c163ccf224c134f`). Claude confirmed `main` at +that commit, signed (`G`), at 14:27 UTC by its own clock. Before the landing, Emil reported one +push refused with GH013, "Required status check "agent review" is expected"; Claude does not know +whether that was before or after 14:20, so the decision does not place it. + +## The landing and the three times — received 2026-11-19 07:51 UTC (08:51 Europe/Berlin) + +> #85 is landed. The head on main is 1bf7e56, merged at 15:25 Europe/Berlin. +> +> Three times for DDD-dec-123, all Europe/Berlin: +> - lifted: +> - restored at 15:28, but with the wrong check name: "agent review (judge)", bound to the gates +> app. That is the workflow job's name, which the gates app never posts, so the rule could not +> have been satisfied by any pull request. +> - corrected at 15:39 to "agent review" from the gates app (integration 5594102), which matches +> .github/repo-standard.yaml. +> +> Record all three. The live rule departed from the written standard from the first time to the +> third, not to the second. Say in the decision that nothing landed between 15:25 and 15:39 +> other than 1bf7e56, and check that against main's history before you write it. +> +> Go ahead with DDD-dec-123 as ruled. + +Claude's notes. + +- **The lift time.** "" is read as the time Emil gave earlier: "lifted 14:20 … UTC" + (Landing facts, above), which is 15:20 Europe/Berlin. UNVERIFIED — Emil review, if he meant + another time. +- **The restore time.** Emil's earlier message said "restored 14:29 UTC" (15:29 Europe/Berlin). + This message says 15:28 and corrects what was restored. The decision follows this message. +- **The check against main's history,** made before writing it into the decision: + - `main` moved once since `84aa466`: a fast-forward to `1bf7e56`, and it is still there at + 2026-11-19 07:51 UTC. + - The only push-triggered run on `main` since 2026-11-10 is validate run 392 (44324064146) on + `1bf7e56`, created 14:25:20 UTC (15:25 Europe/Berlin), which matches the merge time. + - So nothing but `1bf7e56` landed between 15:25 and 15:39, nor since. +- **Agent review runs 89 (14:34–14:35 UTC) and 90 (14:38 UTC)** ran while the live rule required + "agent review (judge)". The rule's name does not change what the workflow runs: both built + main's canon and posted an `agent review` check from the gates App. Neither verdict could have + satisfied that rule; both were `failure`, for want of approval.