* fix (hls): preserve playback reserve across cache eviction
keep a stable session media clock and completed segment timings so cache
eviction and origin rollovers cannot reset playback reserve.
update segment fixtures and cover reserve stability after collection.
* fix (hls): prevent playback stalls during origin manifest retries
mark the origin as degraded on the first retryable manifest failure, cap
retry-after delays, and reevaluate reserve independently of client polling.
use a bounded state commit after an expired splice deadline instead of
leaving the lease live when terminal media cannot be published safely.
* fix (hls): serve custom responses after the live splice deadline
publish a finite lease-bound channel_unavailable playlist on the existing
manifest route when a safe live splice is no longer possible.
renew lease and session activity on successful custom-segment requests
and retain concrete preparation failure diagnostics.
* fix(hls): persist origin qualification for short playlists
Share qualification and retry receipts across leases, preserve them while media publication is pending, and allow controlled recovery after expired acceptance episodes. Prefer eligible regular candidates before short fallbacks.
Add startup, revalidation, replay, cookie and endpoint regression coverage, and document short-window behavior.
Validation during implementation: Linux make test (7009 passed, 23 ignored), focused HLS and endpoint suites, format checks, Clippy and Markdown lint. Offline production library and backend checks passed after the import fix.
* fix(hls): reconcile cache write budgets and revision reclamation
Transfer reserved bytes atomically through staging and publication, preserve cleanup ownership across cancellation, and retire unpinned revisions during local and global reclamation.
Version temp registrations during inventory scans, remove production panic calls, and isolate debug diagnostics from admission results.
Validation: 1,162 HLS tests and 7,038 workspace tests passed; fmt-check, lint, test, markdown-lint and isolated production Clippy passed with dev-slim.
* feat (ai): add local codex files
* fix (hls): replace invalid origin accounts in shared HLS sessions
* fix (hls): synchronize origin account handovers across WebUI viewers
* fix (hls): preserve same-account resume work and backpressure
* fix (hls): resume reduced retries without crossing generations
Allow eligible exhausted acceptance episodes to select reduced retry candidates
while preserving generation, candidate, workload and timing guards. Keep
recovery chains bound to their generation, including changes between selection
and commit.
Add regressions for repaired and persistent resource conflicts, generation
isolation and media delivery through the app's local HTTP origin. Keep shared
manifest fixtures under test/fixtures/hls/reduced-retry.
With startup.mode = optimistic, session.startup stays set for the whole
session lifetime on clear MPEG-TS sources. select_window's startup branch
returned before retain_published_predecessors, so the v3.3.153
three-target-duration retention never ran in that mode. A 3-segment
origin window was served unchanged (e.g. 17.96 s with target 7), both on
retained-session return entries and during steady playback.
Apply the retention after the startup branch's MIN_VISIBLE_SEGMENTS check.
The retention rules are unchanged: only contiguous Ready, renderable,
non-provisioning predecessors are added, never before the previously
rendered first segment, and the window is capped at
TARGET_VISIBLE_SEGMENTS.
* feat(watch): group added/removed contexts per watch entry
- watch entries: regex string or {pattern, on:
[channels|group_added|group_removed]}
- one playlist.watch.changed per refresh with WatchChange list;
playlist.groups.changed aliased
- Telegram/ntfy message length limits
- web UI: Mapping and Watch sub pages, new Mapping icon
- ClusterFlags on create_bitset!, bitflags removed
BREAKING CHANGE: playlist.watch.changed templates must use the new
watch.* fields
* feat(watch): group added/removed contexts per watch entry
- watch entries: regex string or {pattern, on:
[channels|group_added|group_removed]}
- one playlist.watch.changed per refresh with WatchChange list;
playlist.groups.changed aliased
- Telegram/ntfy message length limits
- web UI: Mapping and Watch sub pages, new Mapping icon
- ClusterFlags on create_bitset!, bitflags removed
BREAKING CHANGE: playlist.watch.changed templates must use the new
watch.* fields
* fix hls: require fresh origin window for new playback on retained sessions, count known-length revision bodies as completed
- New playback leases joining a retained shared session no longer start
on
a stale committed window. If the origin window was last confirmed more
than two target durations ago, admission requires a fresh commit
(StaleEntryWindow) using the strict pinned host and no recovery burst.
This prevents the forward jump after fast-start entry.
- RevisionResponseStream tracks bytes sent. A known-length body that is
dropped after its final byte is now recorded as completed and marks
segment completion instead of logging outcome=dropped.
- Add an end-to-end re-entry regression test covering progressive,
first_ready and conservative, plus unit tests for the commit
requirement and revision completion.
* - fix epg: adopt guide channel ID casing on matched live streams
Synchronize the stream's `epg_channel_id` to the exact casing of the
matched XMLTV guide channel. This fixes exact-key B+Tree lookups in short
EPG and client API responses when stream and guide IDs differ in case.
- fix frontend: hide column rail on empty tables
* - fix epg: adopt guide channel ID casing on matched live streams
Synchronize the stream's `epg_channel_id` to the exact casing of the
matched XMLTV guide channel. This fixes exact-key B+Tree lookups in short
EPG and client API responses when stream and guide IDs differ in case.
- fix frontend: hide column rail on empty tables
* fix hls close revision file at reader deadline
Move the progressive revision reader onto its own task and feed the
response stream through a bounded channel. At the deadline the watchdog
aborts the reader, so the file handle and revision pin are released
even if a stalled client never polls the stream again.
Each open revision file now holds its own pin, so GC cannot retire a
revision while a handle is open, including one held by an in-flight
blocking read.
* fix hls close revision file at reader deadline
Move the progressive revision reader onto its own task and feed the
response stream through a bounded channel. At the deadline the watchdog
aborts the reader, so the file handle and revision pin are released
even if a stalled client never polls the stream again.
Each open revision file now holds its own pin, so GC cannot retire a
revision while a handle is open, including one held by an in-flight
blocking read.
---------
Co-authored-by: euzu <euzu@proton.me>
* **New Features**
* Customize table columns by showing, hiding, and reordering them, with keyboard or pointer controls.
* Save layouts per account, restore defaults, and resolve conflicts when preferences change elsewhere. On installations without web authentication, settings are shared.
* CSV tables now support saved layouts for recognized table structures.
* **Bug Fixes**
* Improved pagination bounds and page-size validation.
* Settings requests now return clearer errors for oversized or unreadable request bodies, and recheck access after waiting for settings to become available.
* **Documentation**
* Added guidance for table customization and the settings API.
* fix(frontend): clean up websocket callbacks and reconnect timers on session expiry
* fix(frontend): clean up websocket callbacks and reconnect timers on session expiry
---------
Co-authored-by: euzu <euzu@proton.me>
* fix(m3u): use alias token for Flussonic catch-up archives
Catch-up archive URLs are built from the main account's live URL. On an
alias allocation the M3U URL index only knows live URLs, so the lookup
missed and the archive manifest was fetched with the main stream token.
DVR segments inherited it, and the provider attributed the playback to
the main account.
On an index miss for a Flussonic archive file, resolve the live file in
the same directory through the alias index (one lookup pass for all
candidates) and keep the requested archive file name with the alias
query.
Query-based catch-up templates (append mode) are not covered.
* fix(dvr): retry ffmpeg spawn while executable is busy
Spawning a just-written executable can fail with ETXTBSY when a
concurrently forked process still holds an inherited write descriptor
before its exec. This made run_recording_preempts_fake_ffmpeg_and_
preserves_window_semantics flaky on CI.
Retry the spawn up to five times with a short backoff on
ExecutableFileBusy; other errors fail immediately as before.
* fix(hls): resolve single-variant archive master in shared cache
Archive URLs that return a master playlist with one #EXT-X-STREA
child failed with segment_uri_without_extinf. The variant's media
playlist is now fetched and fed to the cache timeline.
- skip masters with several variants or separate #EXT-X-MEDIA tr
- send entry credentials only to the entry origin
- send master cookies only to the master origin
- merge master and variant session cookies by name
- keep the recovery binding on the entry URL
* **New Features**
* Added an EPG group view with channel counts, programme details, and channel search.
* Added timeline zooming and panning, recording controls for eligible upcoming programmes, and a resizable side panel with keyboard and pointer controls.
* **Improvements**
* EPG group and grid results support JSON or CBOR responses and show up to 1,000 channels per group.
* Channel filters support case-insensitive text and regular expressions; invalid patterns are rejected with an error.
feat(dvr)!: unify live and VOD/series recording into one recoverable DVR
Merge fix/merge_live_and_vod_recording into develop.
- One recording queue for scheduled live captures and VOD/series transfers,
with a shared transition graph, provider-slot scheduling, resumable HTTP
transfers (strong ETag / Last-Modified validation) and ffmpeg live capture.
- Crash-recoverable persistence: B+Tree repository with a journal-first
recovery history; one file can be shared by several library entries.
- Admission checks quota and disk on request, at transfer start (real size)
and before an entry attaches to a finished file; idempotent requests.
- REST is command-only; state reaches clients via owner-filtered WebSocket
snapshots.
- Deleting a recording removes the file once no entry holds it; removing an
entry keeps a finished file and cleans up orphaned partials.
- User ids derive from the configured username (web:/api:); the identity
registry is gone.
- Spanish locale added; ru/ar completed; locale parity test.
BREAKING CHANGE: `video.download` is now `video.recording` (nested
`recording` keys move up, `download_priority`/`recording_priority` become
`priority`); old configs fail to load with a migration hint.
BREAKING CHANGE: `/api/v1/file/download*` and `/file/record` are removed in
favour of `/api/v1/recording/*`.
BREAKING CHANGE: `download.*` and `recording.write` permissions are replaced
by `recording.read|create|manage|delete`; tokens from before the change must
re-authenticate.
BREAKING CHANGE: the old recording queue does not migrate; existing
`identity_registry.json` is ignored.
euzu