feat: daily summary email + skip re-querying unresolvable mis-seeds#2
Merged
Conversation
Add a `maintain` command that runs backfill + resolve in one process (each phase guarded so a failure in one doesn't skip the other) and, with --email, always sends a one-page summary — a daily heartbeat that fires even on an aborted run. Exits non-zero on a phase failure so a supervising timer is marked failed and retries; an SMTP failure does not change the exit code. Add a resolve skip-list: mis-seeds that resolve to no-match or a below-threshold match are recorded (resolve_skip table) and not re-queried on later runs, cutting the repeated Hardcover + Haiku calls for permanently-unresolvable books. The skip key is the normalized title+author, so a book is re-queried automatically once its metadata changes. `resolve --force` re-checks all; `resolve --clear-skips` forgets the list; COLOPHON_RESOLVE_RETRY_DAYS sets an optional re-check TTL (default 0 = never). Bump to 0.2.0; add unit tests for the skip-list logic.
vidaks
added a commit
that referenced
this pull request
Jun 18, 2026
…the lowest book_by_id reported a book's position as bs[0] — the lowest-position membership after order_by position asc. A book in several series has a different position in each: an Expanse novella sits in "The Expanse" (#2.7, the featured series grimmory derives) and "The Expanse (Chronological)" (#0.1). The audit picked 0.1, "containment-matched" its name to grimmory's "The Expanse", and flagged 2.7 != 0.1 — a false number-mismatch. The heal set the (already-canonical) ISBN and refreshed, grimmory re-derived 2.7 from the featured series, and the mismatch came back next run: a heal that re-fired forever without converging. At an hourly cadence the same four books "healed" every run, so every run looked noteworthy and pushed a notification. book_by_id now also returns the featured membership; the audit compares grimmory's number against the FEATURED position — exactly what grimmory derives on refresh. This fixes the four novellas (Drive, The Churn, Butcher of Anderson Station, Spoils of War → number-ok) and, because it uses featured rather than a name match, keeps a genuine mis-seed/variant (grimmory's name disagrees with the featured series, e.g. an Old Man's War book whose featured series is The Human Division) on the name path instead of reclassifying it into a re-firing number-mismatch. Verified on the live library: zero number-mismatch remain; the four oscillating books read number-ok and the maintain run stays silent.
vidaks
added a commit
that referenced
this pull request
Jul 9, 2026
…the lowest book_by_id reported a book's position as bs[0] — the lowest-position membership after order_by position asc. A book in several series has a different position in each: an Expanse novella sits in "The Expanse" (#2.7, the featured series grimmory derives) and "The Expanse (Chronological)" (#0.1). The audit picked 0.1, "containment-matched" its name to grimmory's "The Expanse", and flagged 2.7 != 0.1 — a false number-mismatch. The heal set the (already-canonical) ISBN and refreshed, grimmory re-derived 2.7 from the featured series, and the mismatch came back next run: a heal that re-fired forever without converging. At an hourly cadence the same four books "healed" every run, so every run looked noteworthy and pushed a notification. book_by_id now also returns the featured membership; the audit compares grimmory's number against the FEATURED position — exactly what grimmory derives on refresh. This fixes the four novellas (Drive, The Churn, Butcher of Anderson Station, Spoils of War → number-ok) and, because it uses featured rather than a name match, keeps a genuine mis-seed/variant (grimmory's name disagrees with the featured series, e.g. an Old Man's War book whose featured series is The Human Division) on the name path instead of reclassifying it into a re-firing number-mismatch. Verified on the live library: zero number-mismatch remain; the four oscillating books read number-ok and the maintain run stays silent.
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
1 participant
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.
Summary
Two related changes to the autonomous healer, plus a
0.2.0bump.1. Stop re-querying unresolvable mis-seeds
The nightly
resolvesweep re-asked Hardcover + Haiku about the same permanently-unresolvable books every run (the ~10 NONEs + below-threshold matches), which was the load that tripped the HTTP 429 crash.resolve_skiptable: mis-seeds that resolve to no-match or a below-threshold match are recorded and filtered out of later runs.hash()), so a book is re-queried automatically once its metadata changes — and a heal/lock removes it fromreview-misseedanyway.resolve --force(re-check all),resolve --clear-skips(forget),COLOPHON_RESOLVE_RETRY_DAYS(optional TTL, default0= never).2. Daily summary email
maintaincommand runs backfill + resolve in one process (each phase guarded so a failure in one doesn't skip the other) and, with--email, always sends a one-page summary — a heartbeat that fires even on an aborted run (composed in afinally).SMTP_*/OVERSIGHT_TOenv — no new secret.Tests / checks
unittestcases for the skip-list (fingerprint stability, TTL, store roundtrip, the apply-path skip-writing).compileall+ ruff (CI ruleset) green; no-network render smoke for clean / aborted / cached paths.Deploy
The plexarr
colophon-maintain.sh.j2collapses to a singlemaintain --apply --emailcall (separate repo, deployed after merge).