- Library settings (prefs.json, presets.json, view.json): only a missing
file means "defaults". A file that doesn't parse is kept as
<name>.corrupt-<unix time> before the defaults are used; one that can't
be read (locked, I/O error) is never written this session (save_prefs
refuses, presets/view saves skip it). Warnings go to library.info →
settingsWarnings and Session::take_library_warnings; the UI shows each
once in a window (ui.inspect → notices, button:noticeOk).
- Desktop ui.json (holds the library location): written atomically (temp +
fsync + rename + dir fsync), saved as soon as the library path changes and
every few seconds when it changed, failures reported; a damaged file is
renamed to ui.json.corrupt-<time> and reported, an unreadable one is not
overwritten.
- Quit/close with changes only in memory: retry the save (log, then a
snapshot of everything), and if it still fails cancel the close and ask:
Try Saving Again / Quit Anyway / Cancel (ui.inspect → quitPrompt).
- Tests: engine (damaged prefs kept, unreadable presets/prefs untouched),
desktop (damaged ui.json kept, atomic write), headless UI (quit prompt
buttons, settings notice).
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A GPU render no longer hands back an image it cannot vouch for:
- set the wgpu device-lost callback (wgpu never reports a lost device as an
error, so a driver reset went unnoticed); readback waits with a timeout and
checks the poll result and map callback instead of blocking forever
- each render runs in OutOfMemory/Validation/Internal error scopes; errors
fail that render (OOM and limit failures for that render only, device
errors for the process) and it is redone on the CPU
- every buffer is checked against the device limit when created (no
validation error, no process-wide disable); LIGHTCRAFT_GPU_LIMITS=webgpu|<MiB>
creates the device with lower limits to reproduce small adapters
- the output buffer is cleared and the per-pixel kernel writes alpha 255:
any unwritten pixel (work silently dropped, the issue's all-black export)
fails the render; an all-black result is re-rendered on the CPU
- renders submit per stage and every 16 M invocations, the per-pixel stage
runs in ~4 MP bands: no long submission for a slow iGPU's watchdog to reset
- ui.inspect perf.gpuReason / perf.gpuFallback, app.gpu reason/lastFallback,
System Info and Settings say why the GPU is off (e.g. llvmpipe skipped) and
why the last render fell back
Tests: crates/gpu/tests/fallback.rs (limit, mid-render limit, validation
error, dropped work, black photo), engine export falls back to the CPU image,
banded 3000x2000 GPU/CPU equivalence.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A command whose journal append fails now returns EngineError::NotSaved
("saved in memory but not written to disk: <reason>; LightCraft will retry")
instead of OK. The change stays applied and undoable, its ops stay queued and
are written by the next save (any command, or the frame loop with a 2 s
backoff). Queries and commands that change nothing are unaffected; a failed
compaction is reported in lastError, not as a failed command (and a failed
synchronous snapshot now backs off like a background one).
Session::unsaved(), library.info unsavedOps/unsavedError, ui.inspect unsaved;
the desktop UI toasts the failure and shows a warning on the top bar's cloud
icon until a save succeeds.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
At the 2,000-record threshold the UI thread serialised and fsynced the whole catalog (1-2 s
at 85k photos after streaming). Now `persist()` starts `Journal::snapshot_in_background`:
- the worker writes `catalog.snap` for op N from a clone of the catalog (photos are Arc-shared,
so the clone is cheap) through `Store::background_writer` (a second FsStore handle; MemStore
opt-in; none on wasm/web, which stays synchronous);
- appends go on as before (fsynced before every command returns) and are also kept in memory
as the "tail" (records > N);
- `Journal::poll` (from `persist` and the per-frame `persist_if_dirty`) sees the worker done and
atomically replaces the log with exactly the tail.
Crash points: before the snapshot rename -> old snapshot + whole log; between the rename and the
log rewrite -> new snapshot + whole log (records <= N skipped as stale, existing rule); after ->
new snapshot + tail. A failed snapshot keeps the log whole and retries after another
max_records records (no retry storm). At most one snapshot in flight; a synchronous snapshot
(close, Optimize Library), reopening a library and dropping the Journal first wait for it, so an
older snapshot can never replace a newer one. On-disk format unchanged.
Tests (crates/catalog/src/tests_background.rs, a gated store holds the worker mid-flight and
"crashes" by copying the files): edits during a background snapshot survive crashes at each
point and the trimmed log holds exactly seq N+1.. in order; failure loses nothing and backs
off; sync snapshot waits (no regression); drop waits; real FsStore with a leftover partial
temp file; a proptest over interleavings of edits / start / land / finish / sync snapshot /
crash. Mutations checked: trimming the log to empty, or not waiting in `snapshot`, fail them.
Engine: `threshold_compaction_runs_in_the_background` (crash copy mid-run, reopen without close).
`library.info` -> persistence gains snapshotRunning, failedSnapshots, maxBlockingMs and
lastSnapshot.{background, blockingMs}.
persist_bench, same session (load avg ~160), UI-thread time per threshold compaction:
85k photos: synchronous 0.85-1.7 s (median 1.25 / 1.34 s) -> background: blocking median
45 / 74 ms (clone 1.3-154 ms + log rewrite 9-23 ms); worker 1.2-2.4 s; 230-380
single-op appends during it at median 4-6 ms (same as without)
10k photos: synchronous 72-233 ms -> blocking median 12.5 / 14 ms
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Fixes#11.
Root cause: the grid and the filmstrip remembered the last active photo in egui temp data
with mismatched types (inserted `Option<PhotoId>`, read back as `PhotoId` because of the
`let last: Option<PhotoId> = get_temp(..)` annotation), so the read always missed, "the
active photo changed" was true on every frame and both called `scroll_to_rect` on the
active photo every frame. Any wheel or scroll-bar scroll was undone as soon as the user
stopped (the animated scroll back read as a delayed snap).
- `grid::follow_active` (shared by grid and filmstrip): scroll to the active photo only
when it changed since the strip was last drawn, or when the strip was not drawn on the
previous pass (view switch). Time passing, finished renders, edits and a released scroll
bar no longer move either strip.
- Filmstrip: a vertical mouse wheel scrolls it sideways (horizontal trackpad scrolling
still works); it recentres on a new active photo only when that is off screen, so a
click on a visible thumbnail leaves it alone.
- `ui.inspect` reports `scroll: {grid, filmstrip}`; headless tests in `tests_scroll.rs`.
- docs/parity.md (LR-VIEW-PHOTOGRID, LR-VIEW-FILMSTRIP), docs/control-protocol.md.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
- Engine: export split into prepare_export (session thread, cheap) and PreparedExport::run
(Send: read/decode/render/encode); run_batch with per-photo progress, cancel and collected
errors; export_batch keeps the stop-at-first-error behaviour for scripted calls.
- Desktop: the Export dialog, Export with Preset (menu and context menu) and Export with
Previous run their batch on a worker thread (app.export {background: true}); a progress
panel shows count / file / Cancel; ui.inspect reports export {running, last}.
- Headless: settle waits for a running export; `lightcraft-cli snapshot` finishes it before
exiting. Web keeps foreground exports (downloads are main-thread).
Co-Authored-By: Claude Opus 5.5 <[email protected]>
menubar.rs builds File/Edit/View/Photo/Window/Help from engine + UI
commands (layout table for order and separators, parameterized submenus
for rating, flag, label, sort, add to album), with live labels ("Undo
Exposure"), enabled and checked state; ui.menu.tree exposes it. The
desktop app installs it natively on macOS with muda (pure Rust over
AppKit; no GTK — muda is a macOS-only dependency): app menu (About,
Services, Hide, Quit), Window menu (Minimize/Zoom/Bring All to Front),
Help menu; per-frame state sync; accelerators without Cmd and the text
editing ones step aside while a text field has focus; natively handled
shortcuts are skipped by the egui handler. Other hosts (web, Windows,
Linux) get the same menus in the top bar. Panels/sections/tools move to
the Window menu. Verified with System Events: menus listed with live
enabled state, View > Survey dispatches.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
- ui-egui/softpaint: pure-Rust rasterizer for tessellated egui output (gamma-space premultiplied
blending like egui-wgpu, bilinear texel-centre sampling, scissor clips, tie-break rule so shared
edges blend once, parallel row bands, deterministic) + CPU texture mirror from TexturesDelta.
- ui-egui/headless: HeadlessView (offscreen context with our fonts/theme) and Headless (windowless
app session: logic+ui per frame, answers viewport Screenshot/InnerSize/Close, settle loop,
control requests through the shared handler).
- ui.screenshot {headless: true} in the running app renders from the logic tick (works while the
window is occluded / display asleep); windowed captures fall back to headless after 2 s.
Screenshots now wait for in-flight renders, not just queued ones.
- lightcraft-cli snapshot [--demo|--library DIR] [--script f.jsonl] -o out.png --size WxH --scale S.
- Fixes found with it: dialog backdrop was painted over the dialog itself (now an area below the
window); dialog content padding; crop overlay options use our buttons (registered widget ids).
- Docs: control-protocol.md (headless section), AGENTS.md (prefer headless when unattended).
Co-Authored-By: Claude Opus 5.5 <[email protected]>
crates/mcp (lightcraft-mcp, L5): hand-written MCP over stdio (JSON-RPC 2.0,
protocol 2025-06-18 with 2025-03-26/2024-11-05 negotiation): initialize,
ping, tools/list, tools/call, resources/list/read. Two backends behind one
trait: Headless (in-process engine Session; answers engine.execute,
engine.commands, ui.render, app.export) and Remote (the desktop app's
loopback JSON-lines control channel; lazy connect + reconnect).
Tools: helpers (list_commands, run_command, import, query_photos,
select_photos, list_controls, get/set_develop, apply_preset, crop,
render_photo -> image content, export), UI tools when connected
(screenshot -> image content, inspect_ui, set_ui, list_widgets, click,
press_key, type_text, pointer_gesture), plus one cmd_<id> tool generated
per command in the registry (--compact omits them).
apps/lightcraft-cli: `mcp [--connect [ADDR]] [--demo] [--compact] [FILES]`,
`render <in> -o <out> [--set control=value]... [--settings] [--preset]
[--size] [--quality]`, `commands`, `controls`.
Acceptance e2e: crates/mcp/tests/e2e.rs sets exposure over MCP and renders
(headless and through the TCP transport); apps/lightcraft-cli/tests/cli.rs
does the same through the real binary's stdio. Docs: docs/mcp.md,
docs/control-protocol.md. Also rustfmt of one line in pipeline/finish.rs
(main was not fmt-clean).
Co-Authored-By: Claude Opus 5.5 <[email protected]>
