Subscriptions: owner-first matching, idle wake, shared HMAC control contract (KSE-0199) #23
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "devex/steer-subscriptions-impl"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Address founder review of PR #21 (Redmine #46):
candidates and the predicate re-check to owner_identity = envelope
addressee; set-based re-check pins each OR branch to its own
subscription id so one match cannot select every row of the owner.
preserving the primary payload agent_originated verbatim; relay,
self-authorship, and pending-copy coalescing guards bound wakes.
the schedule control-plane contract (dispatch PR #22) - hub-resolved
owner, HMAC-SHA256 over stamp/path/body, 5min/30s skew, loopback-only,
fail-closed 403s. Replaces the injectable caller-resolver seam.
re-checked after insert with self-rollback.
Tests: HMAC control auth table, CRUD round-trip, cross-identity
isolation, tampered-body rejection; match/dedup/coalescing/trust
suites green. Full package shows only the two pre-existing base
failures (Claude client, org TOML runtime), reproduced on clean HEAD.
Review of exact head
edf3aa7ad1. Ownership-before-filter and the authenticated control direction are improvements, but these reliability issues still need correction before deployment:Run SQLite plus actual PostgreSQL tests through the authoritative Nix package check. Root can run this on your final signed head; the current main key.store flake already provisions an isolated PG17 test cluster (KEYSTORE_TEST_POSTGRES_URL). Preserve owner access isolation, source provenance and no self/relay wakes. Do not ship a docs-only response; update signed implementation and its KSE/hub integration.
Correction to review4731 point6: firstNonEmptyString already discards the sentinel, so absent Kind/Summary fields do fall back correctly. That claimed bug is withdrawn. The stale caller-resolver comments still need cleanup. Replay/coalescing/quota/duplicate-wake and current-main ancestry findings remain.
Canonical Nix regression proof on exact
edf3aa7head (isolated SQLite and PG17, no production queue/model/SMS used): derivation /nix/store/fh9phcz267q9nfrjzdgs3zyakyrqlv3k-agent-dispatcher-0.1.0.drv failed its checkPhase as expected.The check used the current key.store flake's agent-dispatcher package with only src and the selected test pattern overridden; PG was build-local. Full log and regression source: /root/work/codex-subscriptions-regression/{nix-review-regressions.log,internal/agentdispatcher/root_review_test.go}. Please include these cases in the final production test suite and run the complete authoritative suite after repairs. The separate point6 nil-fallback claim was withdrawn in comment5877.
edf3aa7ad1to913a8fe293devex referenced this pull request2026-09-14 03:43:09 +00:00
Superseded by merged PR25, retaining this PR's signed DevEx history plus the reviewed durable wake repair. Deployment and live acceptance remain tracked in #46 / key.store PR283.
Pull request closed