ADR-028 started persisting videos.duration_s (previously fetched-then-discarded by the ADR-023 filter). New rows get it on discovery; existing rows stay NULL until their next discovery pass re-upserts them (the channel_title backfill stance, migration 014).
Consequence: the burst's length-aware selection (OnboardBurstVideoIDs) and any future length-aware feature (feed sort, "long read" badge) degrade-open to "unknown" for the back-catalogue until it churns through discovery. At pilot scale the two existing users' ~750 rows are all NULL today.
This is recorded as deliberate in ADR-028 (lazy is fine for a connect-time burst, which only fires for new users whose discovery runs first). File is to decide later, not now:
Accept lazy backfill (close), OR
One-time backfill: a maintenance pass that enriches existing rows via batched videos.list (50 ids/call, 1 quota unit — cheap, not the timedtext throttle).
ADR-028 started persisting `videos.duration_s` (previously fetched-then-discarded by the ADR-023 filter). New rows get it on discovery; **existing rows stay NULL until their next discovery pass re-upserts them** (the channel_title backfill stance, migration 014).
Consequence: the burst's length-aware selection (`OnboardBurstVideoIDs`) and any future length-aware feature (feed sort, "long read" badge) degrade-open to "unknown" for the back-catalogue until it churns through discovery. At pilot scale the two existing users' ~750 rows are all NULL today.
This is **recorded as deliberate** in ADR-028 (lazy is fine for a connect-time burst, which only fires for *new* users whose discovery runs first). File is to decide later, not now:
- [ ] Accept lazy backfill (close), OR
- [ ] One-time backfill: a maintenance pass that enriches existing rows via batched `videos.list` (50 ids/call, 1 quota unit — cheap, not the timedtext throttle).
Source: ADR-028, `docs/specs/onboarding-wow-burst.md` §1.
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
ADR-028 started persisting
videos.duration_s(previously fetched-then-discarded by the ADR-023 filter). New rows get it on discovery; existing rows stay NULL until their next discovery pass re-upserts them (the channel_title backfill stance, migration 014).Consequence: the burst's length-aware selection (
OnboardBurstVideoIDs) and any future length-aware feature (feed sort, "long read" badge) degrade-open to "unknown" for the back-catalogue until it churns through discovery. At pilot scale the two existing users' ~750 rows are all NULL today.This is recorded as deliberate in ADR-028 (lazy is fine for a connect-time burst, which only fires for new users whose discovery runs first). File is to decide later, not now:
videos.list(50 ids/call, 1 quota unit — cheap, not the timedtext throttle).Source: ADR-028,
docs/specs/onboarding-wow-burst.md§1.