v0.3.128
Fixed
- Reverted v0.3.121's "skip backend image fetch for TMDB CDN URLs" optimisation. It traded a one-time download cost on resolve for a per-page-load browser fetch — cache reads stopped being instant because the browser had to fetch from
image.tmdb.orgevery time. On filtered Scanner views with hundreds of cards, the browser's per-host connection limit + TMDB CDN variability made posters trickle in or never load. Backend now downloads + base64-caches every poster on resolve so cache reads render instantly, the way they did pre-v0.3.121. - Lazy backfill for URL-only cache rows written during v0.3.121-127: on cache read, any row with
sourceoftmdb/tmdb-manualwhoseposter_urlis still a CDN URL (not adata:URI) gets its image downloaded inline (bounded concurrency 8) and the cache row updated. One-time-per-row hit; subsequent reads are instant.
Docker images
Pinned to this release:
docker pull ghcr.io/i-ial9000/shrinkerr:0.3.128 # portable (amd64 + arm64, CPU)
docker pull ghcr.io/i-ial9000/shrinkerr:0.3.128-nvenc # x86_64 + NVIDIA GPU (NVENC)
docker pull ghcr.io/i-ial9000/shrinkerr:0.3.128-edge # as above but ffmpeg master
docker pull ghcr.io/i-ial9000/shrinkerr:0.3.128-edge-nvencFloating tags (:latest, :edge, :nvenc, :edge-nvenc) track the most recent release of each lineage.