Parity Certification Report¶

Scope¶

This report captures Torrentarr's parity status after rebasing the audit from qBitrr v5.12.10 through v5.14.4-1. Torrentarr schema is 6.14.4.

Primary tracking artifacts:

• docs/parity/full-parity-matrix.md
• docs/parity/full-parity-path-report.md
• docs/parity/contract-baseline.md
• docs/parity/contributor-reference.md
• docs/parity/overview.md

Implemented in This Pass¶

Phase 0 — Baseline rebasing¶

• Rebased the parity audit from qBitrr v5.12.10 to v5.14.3-1.
• Reclassified parity rows that were still carrying the prior closeout's full claim without latest-main verification.

Phase 1 — Critical correctness¶

• Import completion parity (5.12.7): TorrentProcessor no longer marks torrents imported when the Arr scan is merely queued. It now waits for IArrImportService.IsImportedAsync() to confirm the item has left Arr's queue before persisting Imported = true and applying the imported tag / AutoDelete follow-up.

Implemented later (qBitrr 5.14.3-1 path-report closeout)¶

• Post-import FFprobe AutoDelete cleanup (folder_cleanup), Host DI, optional FFprobeAutoUpdate.
• SearchMissing master switch for search loop and Ombi/Overseerr request marking.
• Explicit loop_completed (reset Searched and Upgrade).
• Lidarr QualityMet from percentOfTracks + cutoff/upgradeAllowed.
• Removed unused IsQualityUpgradeAsync.
• In-process dual torrent/search tasks in ArrWorkerManager (Host does not spawn Torrentarr.Workers).
• ExpectedConfigVersion / default config references aligned to 6.14.3.
• Readarr ([Readarr-*]), AutoUpdateChannel, fractional durations, qBit init retry, seeding -1 merge, and existing-DB table creation via ManualSqliteMigrations.
• config.example.toml now surfaces MatchSubcategories in the primary qBit example for parity with upstream config docs.
• OpenAPI helper scripts and contributor docs now point at latest qBitrr master by default instead of the old 5.12.3 pin.

Implemented later (qBitrr 5.14.4-1)¶

• Stalled-upload MaxSeedingTime clock (stalledUP observation, not last_activity); HnR still uses real seeding_time / ratio.
• Per-series Sonarr episode HTTP skip (including 415); transport abort; all-fail rethrow so ingest is not marked complete.
• ExpectedConfigVersion / default config references aligned to 6.14.4.

Validation Evidence¶

Backend tests (dotnet test --filter "Category!=Live"):

Frontend tests (cd webui && npx vitest run): not re-run this pass (no frontend changes). Prior closeout: exit code 0 (167 tests).

OpenAPI drift should now be run against latest upstream master (or an explicit QBITRR_OPENAPI_REF override) after regenerating docs/assets/openapi.json.

Focused regression checks added/updated:

• TorrentProcessorTests / TorrentProcessorAutoDeleteTests — post-import AutoDelete FFprobe cleanup; import remains pending until Arr confirms completion
• ArrWorkerManagerTests — SearchMissing master switch, explicit loop_completed reset of Searched+Upgrade, in-process dual loops
• ArrSyncServiceTests — Lidarr QualityMet from percentOfTracks + cutoff/upgradeAllowed
• MediaValidationServiceTests — missing ffprobe treated as valid; ebook skip; allowlist

Matrix Status¶

Latest-main user-facing parity is closed against qBitrr v5.14.4-1. Remaining intentional-divergence rows in full-parity-matrix.md are architecture or packaging only (process isolation, WAL vs db_lock, fork session sharing, Pathos, placeholder defaultdicts, setup.py / CI autofix).