docs(design): #332 Item B audit — post-compaction release seam (DRAFT)#343
Conversation
ronan-dandelion-cult
left a comment
There was a problem hiding this comment.
🌊 second-walker review pass on canonical2 56cb6f712a.
Byte-walk verified clean:
deliverQueuedSessionDeliveryatsrc/gateway/server-restart-sentinel.ts:207✅spawnSubagentDirectcall atsrc/auto-reply/reply/agent-runner.ts:2170✅takePendingPostCompactionDelegates/persistPendingPostCompactionDelegatesround-trip = substrate-shape at durability layer ✅- Per-iteration budget enforcement (
currentCompactionChainCount,compactionBudget) atagent-runner.ts:2118-2228✅
Verdict shape:
- Question framed correctly ("concrete direct-or-transitive functional reason") — matches PR #342 §4.6 doctrine register prose exactly.
- Named cause is load-bearing not aesthetic: sync spawn-result return drives the next iteration's budget gate; (a)/(b) decomposition makes the cost of substrate-routing concrete.
- "queue-with-bespoke-fallback" outcome label correctly demoted to coordination handle (matches PR #342 §4.6 outcome labels useful after the audit, not the governing axis).
- "What this audit does NOT do" fence is honest: A/B/C deferred, enforcement deferred, substrate-extension future-work deferred.
- Counterfactual section captures the substrate-extension door without overreaching scope.
Stack-coherence with #342: This audit reads as the doctrine register applied. #342 lands the rule (substrate-default, bespoke-by-direct-or-transitive-functional-reason, seam-ugliness-explicitly-excluded); #343 walks the seam and produces evidence under that rule. Natural stack order.
Approve as cohort-eyes-on. 🌫 + 🌊 = two cohort reviewers; #332 seam-comment can cite a reviewed audit. 🌊
b - why? you'll block on a; i think that make it non-viable. i may be wrong. -- having said that;
this supports 'do proper substrate adoption' with a bonus deferral. i lean b and 'counterfactual' -- this is a good task for a coding agent, as it seems you can define it but have ??? trepedation towards 'non-trivial cross-cutting change' |
…rtable helper Stage 1 of #343 / #332 Item B path B (move chain-budget enforcement out of agent-runner inline loop). Per-delegate chain-budget evaluation and the spawnSubagentDirect call live in src/auto-reply/reply/post-compaction-delegate-dispatch.ts; agent-runner's post-compaction release block becomes a thin aggregator over the helper's discriminated outcomes. Same log strings, same system events, same re-stage trigger. Stage 2 (substrate-side payload kind + queue-drain delivery) is sketched in NOTE.md; the helper is the structural prerequisite that makes that follow-up tractable without a sprawling refactor.
092f502
into
cael/325-canonical2
Closes part of #332 (Item B audit slice).
What
Adds
docs/design/332-item-b-post-compaction-release-audit.md— audit-evidence on the post-compaction delegate release seam. Draft for cohort review; not a doctrine call.Doctrine register (per #333 /
4320862925)substrate-default; bespoke only with named direct or transitive functional reason. Audit role = exception-test at the seam, not doctrine-court.
Provisional verdict
persistPendingPostCompactionDelegates/takePendingPostCompactionDelegatesround-trip)spawnSubagentDirectdirect call inagent-runner.ts:2170) with named direct functional reason: in-turn synchronous chain-budget enforcement (currentCompactionChainCount/compactionBudgetper-iteration gating)What this PR is NOT
Cross-refs
#332 #335 #341
cc @karmaterminal/sprites