v0.9.68

Changed

• Language detection no longer masquerades as a native-language source. It set language_source='detected', conflating "detection ran" with the native language's provenance — which hid that the native was still a heuristic guess and excluded the title from the heuristic→API refresh. Split into a separate tracks_detected flag; language_source now means only native provenance (api/heuristic/manual). Detection also no longer downgrades a TMDB- or user-set native to a heuristic guess. Existing detected rows migrate to heuristic + tracks_detected=1.

Fixed

• A cleanup job that removes both audio and subtitle tracks is now labelled "Audio + Sub cleanup" instead of "Remux." The both-removals case fell through to the remux label.

Docker images

Pinned to this release:

docker pull ghcr.io/i-ial9000/shrinkerr:v0.9.68         # portable (amd64 + arm64, CPU)
docker pull ghcr.io/i-ial9000/shrinkerr:v0.9.68-nvenc   # x86_64 + NVIDIA GPU (NVENC)
docker pull ghcr.io/i-ial9000/shrinkerr:v0.9.68-edge    # as above but ffmpeg master
docker pull ghcr.io/i-ial9000/shrinkerr:v0.9.68-edge-nvenc

Floating tags (:latest, :edge, :nvenc, :edge-nvenc) track the most recent release of each lineage.