immich-folio/CHANGELOG.md

Changelog

All notable changes to this project are documented here.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning. Releases up to and including v0.9.2 are documented in the GitHub releases.

[0.10.0] — 2026-08-10

Added

• Photo Essay mode — storytelling pages with an RSC markdown parser and fullbleed/pair layouts, plus a visual essay block builder in the admin panel.
• Client proofing — favorite selection with compact URL bitmasking and an export modal.
• Subpage enable toggle — deactivate a subpage without deleting it.
• SEO — configurable subpage title template, metadata descriptions, and generateMetadata on the about page.
• New theme preset studio-modern. The Leica language of studio rebuilt around the precision grotesque Archivo, with IBM Plex Mono for every piece of photographic metadata. Hairline rules, zero radius, an indexed hero navigation with album counts, an always-visible caption bar under album covers, and a film-edge EXIF strip in the lightbox. Selectable in the admin panel, the dev toolbar, and content/settings.yaml. See docs/theming.md.

Security

• GET /api/admin/analytics now requires an admin session.
• Docker images are scanned with Trivy on every publish; results land in the GitHub Security tab.

Changed

• Album sort order now mirrors the Immich timeline (#350). Albums were sorted by fileCreatedAt, the capture instant in UTC. Immich sorts its timeline primarily by localDateTime — the capture time in the photographer's local zone — and uses fileCreatedAt only as a tie-breaker. The two keys are identical for the vast majority of albums; they diverge for albums spanning time zones, where the order shown in Immich Folio no longer matched the Immich UI. The sort keys were changed accordingly, and unparseable dates now fall back to 0 instead of NaN (a NaN comparator makes sort() free to return any permutation).

Fixed

• Cache staleness — ?fresh=1 bypass, an admin diagnostic banner for albums Immich returns empty, and revalidation after admin saves.
• Admin status panel reported two faults that were not faults (#341). settings.yaml was treated as mandatory although getConfig() falls back to defaults without it, so anyone running with only a gallery.yaml saw a permanent "Config Integrity: Degraded". Only a file that exists and cannot be parsed counts as degraded now. A failed status request was also rendered as if it were a result — an expired admin session looked identical to a real outage — so there is now a third state for "the check did not run".
• Lightbox showed its photo counter twice. The counter renders a screen-reader label ("Photo 3 of 53") alongside the compact display ("3 / 53"), but .sr-only was not defined in any stylesheet, so both strings were visible on top of each other.
• Docker — health check targets 127.0.0.1 instead of localhost, base image moved to node:22-alpine.
• Docs — the "nature / travel journal" example in docs/theming.md used a preset named botanica that does not exist; copying it produced an Unknown theme preset error. It now uses editorial.
• Map page ignored the theme. app/map/map.css referenced custom properties that never existed (--font-heading, --font-body, --border-color, --radius), so the map title, popups, and container fell back to browser defaults instead of the configured fonts, borders, and radius. This affects every preset — the map now picks up the theme like the rest of the site.

Upgrade notes

No migration is required. No configuration schema changed, and the six existing presets render exactly as before. Pull the new image, restart, done.

Rolling back after switching presets needs one edit. If you select "Studio Modern" and later downgrade to a version that predates it, the theme.preset: studio-modern left in content/settings.yaml is a preset the older version does not know, and config loading fails with:

Unknown theme preset "studio-modern". Valid presets: studio, minimal, editorial, classic, noir, monograph

Set theme.preset back to one of the listed presets in content/settings.yaml before downgrading. /admin stays reachable if you hit this, so the change can also be made after the fact.

If you copied docker-compose.override.yml.example, update your health check by hand. The example now uses wget -O /dev/null instead of --spider, but your own docker-compose.override.yml is not tracked by git, so pulling the new version does not touch it — and a health check defined there overrides the one baked into the image. --spider does not read the response body and disconnects mid-RSC-stream; when that abort coincides with a slow render after CACHE_TTL expiry, Node tears down the TransformStream controller while Next.js is still writing into it:

TypeError: controller[kState].transformAlgorithm is not a function

Replace "--spider" with "-O", "/dev/null" in your override file:

healthcheck:
  test: ['CMD', 'wget', '--no-verbose', '--tries=1', '-O', '/dev/null', 'http://127.0.0.1:7211/']

A full GET checks the same thing without the race. Installations without an own health check need no action — the image ships the fixed one.