# Aperture 1.0.1 Aperture 1.0.1 brings the changes from dev.1 through dev.4 to the stable channel. It reduces the work Aperture asks Emby to perform, corrects job countdowns and schedule editing, improves search and recommendations, and adds Encore response styles. ## Emby and background jobs - Recommendation libraries keep unchanged media links, metadata, and artwork in place. Repeated publishing skips unnecessary library refreshes, reducing repeated scans and media probes. - Emby-heavy jobs take turns, with shared request limits, timeouts, and paced library refreshes. Aperture manages scanning for its generated recommendation libraries and preserves source-library settings. - Job cards calculate **Next Run** using the server's active schedule, minute offset, and timezone. Jobs waiting behind other Emby work show **Queued**. - Turning a job on or off preserves its schedule. The schedule dialog lets you edit minute offsets and shows the server timezone. - Catalog sync retries failed episode pages and checks completeness before removing stale records. A controlled 23-minute live test completed two movie-and-TV publishing passes without failed publishing or browsing requests. The repeat pass triggered no media probes or library refreshes. Longer observation is needed to assess the previously reported Emby stalls. ## Search, recommendations, and requests - Search handles punctuation and alternate titles more reliably, ranks title matches first, and shows title results while semantic results load. - Search, Similar titles, and graph exploration check the viewer's current Emby access. Similar results honor the configured watched preference. - Recommendation profiles handle embedding scale and model changes more consistently. Highly rated titles and saved interests can provide starter picks when watch history is sparse. - Gap Analysis retains the last successful movie scan and can check released TV seasons and episodes against the Emby inventory. - Seerr requests recheck availability and request status, prevent overlapping submissions, and reconcile uncertain responses. ## Encore Choose **Quick**, **Standard**, or **Expanded** for the next reply, and set a default in AI Settings. Supported models use their native reasoning controls; other models use response instructions and tool limits. A reply-specific choice resets after sending. Encore checks exact library details and release candidates against Emby. The cost estimator includes supported model prices, cached-input rates, and recent measured usage. Private provider continuation data stays separate from the visible conversation. ## Administration and updating Restores stop if the requested safety backup fails and apply the restored SQL in a transaction. Allow temporary disk space for decompressed backups. Both release architectures pass startup, database-migration, and health checks before their image digests are published. Upgrading from **1.0.0** applies migrations **0124–0126** for Encore continuation data, usage aggregates, and legacy job staggering. The schedule migration fills unset legacy minute offsets and preserves explicit values, including zero. Installations already running dev.4 have no additional migrations. 1. Back up PostgreSQL and uploaded artwork, keeping the existing database and mounts. 2. Use `ghcr.io/dgruhin-hrizn/aperture:latest`, or pin `ghcr.io/dgruhin-hrizn/aperture:1.0.1`, then pull and recreate the Aperture container. 3. Confirm `/health` reports **1.0.1** and a connected database. Refresh the Aperture page to load the updated interface. The release includes the standalone `aperture-mcp.mjs` bridge and its SHA-256 checksum. The stable Docker tags are `1.0.1`, `1.0`, and `latest`, with images for Linux AMD64 and ARM64. [Full comparison with 1.0.0](https://github.com/dgruhin-hrizn/aperture/compare/v1.0.0...v1.0.1)