Releases: binhex/gamarr
Releases · binhex/gamarr
Release list
v2.0.1
v2.0.0
v1.2.0
What's Changed\n\n- Download tracking visibility — post-processing summary now shows in-progress downloads count\n- Noise eliminated — per-game no-op messages at DEBUG instead of INFO\n- Next-run scheduling — post-processing announces next run time at INFO\n- Interval corruption fixed — acquisition listener no longer corrupts post-processing interval\n- Return type cleanup — _process_one returns str|None, phase functions return bool\n- 789 tests passing, ruff/mypy/pre-commit all clean\n\nFull Changelog: v1.1.17...v1.2.0
v1.1.17
v1.1.16
What's Changed
- fix: decode HTML entities in FreeGOG A-Z page titles — HTML entities like
&\#039;in game titles were not decoded before cleaning, causing normalisation mismatches ("baldur039sgate3" vs "baldursgate3") that silently prevented source matching for any game with entities in its title (e.g., Baldur's Gate 3) - backlog progress tracking — new BacklogProgress table enables per-year page advancement across cycles instead of always restarting from page 1. Year scanning range now computed from max_pages. Added cycles-remaining display in Phase 5 (
Backlog progress: X of Y pages...) - fix: library scanner indexes parent dirs for version subdirectories (v1.1.15)
- fix: max_cycle_pages now correctly limits per-cycle scan depth (v1.1.14)
Full Changelog: v1.1.14...v1.1.16
v1.1.15
What's Changed
- fix: library scanner now indexes parent dirs — Game directories with version subfolders (e.g.
Fallout 3/v1.7.0.3_(12034)/Fallout3.exe) now correctly match library checks. The scanner walks up from game-file directories and indexes ancestor names. - fix: FreeGOG keyword-check guard —
_skip_for_reject_keywordshelper gates the keyword check to FitGirl only, preventing FitGirl-specific HTTP fetches and warning spam on FreeGOG matches. - fix: max_cycle_pages now correctly limits per-cycle scan depth (v1.1.14)
Full Changelog: v1.1.13...v1.1.15
v1.1.14
What's Changed
- fix: max_cycle_pages now correctly limits per-cycle scan depth — The per-cycle page limit was stored but never read; cfg.max_pages was passed instead, causing entire backlog to scan in one cycle
- switch to year searching, remove week restrictions
- fix: backlog scan now resumes from stored page instead of re-scanning from page 1
- docs: max_weeks to max_pages config rename
- docs: config enum validation design spec
- docs: year-specific Metacritic browsing
Full Changelog: v1.1.12...v1.1.14
v1.1.13
What's Changed
- fix: backlog scan now resumes from stored page instead of re-scanning from page 1
Full Changelog: v1.1.12...v1.1.13
v1.1.12
What's Changed
- fix: route APScheduler logging through Loguru format
Full Changelog: v1.1.11...v1.1.12
v1.1.11
What's Changed
- feat: colour-highlighted phase banners + standardised source name capitalisation
- docs: log phase colour and source name caps implementation plan
- docs: log phase colour and source name caps design spec
- feat: phase-banner logging with backlog heartbeat and rewritten messages
- docs: logging overhaul implementation plan
- docs: logging overhaul design spec
Full Changelog: v1.1.9...v1.1.11