Fixed
- Bounded automatic-deletion recording favors actionable and rejected decisions over waiting/disabled candidates. Operation retention preserves each target's latest final outcome and automatic decision before intermediate events.
- Operation diagnostics record late favorites and sibling-protection skips with explicit reasons. A generic failure diagnosis no longer replaces an already-recorded guard outcome.
- Future duplicate and upgrade-leftover deletions retain operation checks, provider/file observations and final outcomes after their target rows disappear, without inventing cleanup candidate IDs. Recording failures leave the operation result unchanged.
- Playback diagnostics include stored activity and username-completeness flags, field-specific unavailability reasons and safe event identity-conflict indicators.
- Duplicate and upgrade-leftover job lookup resolves stored target associations, and relevant jobs appear in media timelines even before operation tracking starts.
- Diagnostics export favorites controls, aliased requester-watch mappings, ARR fallback settings and complete quality-profile action targets, preserving identity relationships across settings and rule evidence.
- Media diagnostics include sibling protections and requests that can block automatic cleanup, stored resync/unavailable-rule holds, manual deletion schedules and deletion-request playback deferrals.
- Support timelines preserve the latest final outcome and automatic-deletion decision for each target/run before sampling intermediate failures, so older errors cannot hide a successful recovery.
- Support bundles include stored background-job result counters, task identities and categorized failures, including failures before task tracking starts. Recent job samples preserve each task's latest job and failure.
- Support bundles retain the latest run of every task, so frequent playback refreshes cannot hide cleanup task runs. Stored playback deferrals and recorded automatic-deletion decisions explain why a completed task left a candidate in place.
- Cleanup scans preserve candidates' original flag dates and deletion timers when a previously matched rule is skipped or its data is temporarily unavailable. Automatic deletion stays on hold until a successful scan verifies the candidate; confirmed non-matches are still removed. #526
- Cron task schedules now use standard weekday numbering:
0and7are Sunday, and1-5is Monday through Friday. Weekly defaults and presets now use weekday names. Invalid cron expressions are rejected before saving, including for disabled tasks. #525
Changed
- Support bundles can optionally attach up to five already-completed diagnostic checks, preserving their timestamps and scope without rerunning them. Check snapshots are cached briefly for the requesting administrator and respect path-sharing controls.
- Candidate and deletion support bundles select all diagnostic sections by default and show automatic-deletion run totals, revalidation evidence and missing-history limitations. Preparing a bundle does not rerun rules or retry deletion.
- Upgrade note: Existing cron schedules with numeric weekdays now run on the corrected days. Review custom task schedules after upgrading, especially automatic deletion schedules. Saved expressions are preserved; untouched weekly defaults switch to named weekdays.
- Updated fastapi and platformdirs.
What's Changed
Full Changelog: 0.6.12...0.6.13