Pipeline admission: queue entire epics, delivered in order#1793
Open
psdjungpulzze wants to merge 4 commits into
Open
Pipeline admission: queue entire epics, delivered in order#1793psdjungpulzze wants to merge 4 commits into
psdjungpulzze wants to merge 4 commits into
Conversation
…ce-release escape hatch S3 goal dependency gate: flip GATING_INCOMING_BLOCKS_EDGE so a CANCELLED blocks-predecessor keeps gating its successors (removes the `status != cancelled` carve-out that silently released dependents onto code that never landed). Preserves S2: a merged predecessor still releases (`prState != merged`), and a non-boundary aggregation-parent still releases on accept. Ship the escape hatch in the same change so a chain can never deadlock: - add `forceReleasedAt` / `forceReleasedBy` to EngineGoalEdge (additive migration); - `forceReleasedAt: null` becomes a top-level clause of GATING_INCOMING_BLOCKS_EDGE, so a force-released edge drops out of every gate seam (queue.ts candidacy, countGatingGoalBlockers, the fleet-queue surfacing query, starvation.ts) — all share the constant; - `forceReleaseGoalDependency(edgeId, releasedByUserId?)` stamps the edge (preserved for audit, distinct from removal) — idempotent, throws EngineNotFoundError on miss; - POST .../goals/[goalId]/edges/[edgeId]/force-release, authorizeProjectWrite + the same incident-edge / in-project guard as the DELETE edge route. Tests: updated the exact-shape assertions (goal-edge + queue), inverted the cancelled-releases test to cancelled-HOLDS, added force-release release + function + route (success/not-found/wrong-project) coverage. No agent-config impact: this adds an operator HTTP route, not a registered agent tool, and no new agent-visible entity/workflow. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Review returned the goal for patch coverage below threshold (78.9%): the catch arms added with the force-release escape hatch were unexercised. - goal-edge.test.ts: forceReleaseGoalDependency rethrows a non-P2025 db error unchanged (covers the non-"missing edge" rethrow). - force-release route.test.ts: a typed EngineNotFoundError from the stamp maps to a clean 404 (catch → engineErrorToHttp → mapped response), and a non-engine error rethrows to a 500 (the `throw e` arm). All changed lines in route.ts + forceReleaseGoalDependency now covered. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…force-release control S3 visibility & UI for the cancelled-predecessor HOLD (backend in e73b6b0): 1) Fleet-queue surfacing: the `waiting_on_deps` blockedBy detail select no longer filters predecessors to non-terminal status (which HID a cancelled blocker behind the generic "Waiting on deps" label). It now reuses the full GATING_INCOMING_BLOCKS_EDGE.fromGoal predicate (+ project scope), so a CANCELLED gating predecessor IS surfaced — each blockedBy entry carries `cancelled` + a `reason` ("predecessor cancelled — held pending decision"). Gate classification is unchanged (the `incomingGoalEdges.some` still uses the full predicate). 2) Goal-detail gate cue: flipped so a cancelled (non-force-released) incoming blocks-predecessor reads as STILL GATING, matching the real gate. listGoalEdges now carries `forceReleasedAt` so a force-released edge reads as released; the cue is `forceReleasedAt == null && status != accepted` (best-effort display, claim path stays authoritative). 3) Operator control: the dependency manager surfaces the held state + reason on a held incoming edge and adds an explicit Force-release pill (amber, Interactor design system) wired to POST .../edges/[edgeId]/force-release. After release the edge shows as "Force-released" (control hidden) and the gate cue clears on refetch. The remove-edge control remains the alternate escape hatch. No agent-config impact: operator HTTP route + UI only; no new agent-visible entity/tool/workflow. Tests: fleet-queue cancelled-surfacing, goal-detail cancelled-holds + force-release released cue, goal-edge forceReleasedAt passthrough, UI held-reason + force-release control + released state. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
S3 — A cancelled mid-chain predecessor HOLDS its successors (no auto-release)
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
3 participants
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.
No description provided.