Fixed
- Watcher hard-DELETEd rows under temporarily-unmounted media volumes. The watcher's stale-row pass computed
stale = known_paths - disk_filesacross all configured media_dirs as one set. When one media_dir was missing from disk (e.g. running a docker-compose with a subset of volumes for RC testing), every row under it was flagged stale and deleted — recoverable only by a full rescan. v0.7.7 scopes the stale set to media_dirs that exist on disk this cycle, matching the scanner's existingcompleted_pathsrule. Adds a sanity belt that aborts stale-removal if >50% of walked-dir rows would be deleted in one cycle. Loudly logs both events.
Docker images
Pinned to this release:
docker pull ghcr.io/i-ial9000/shrinkerr:0.7.7 # portable (amd64 + arm64, CPU)
docker pull ghcr.io/i-ial9000/shrinkerr:0.7.7-nvenc # x86_64 + NVIDIA GPU (NVENC)
docker pull ghcr.io/i-ial9000/shrinkerr:0.7.7-edge # as above but ffmpeg master
docker pull ghcr.io/i-ial9000/shrinkerr:0.7.7-edge-nvencFloating tags (:latest, :edge, :nvenc, :edge-nvenc) track the most recent release of each lineage.