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 <noreply@anthropic.com>
Local browse records can be most of a large catalog (85k photos, 90 %
Local: 313.5 MB snapshot). Untouched Local records of folders whose last
browse is older than library.preferences forgetLocalDays (default 30,
0 = never) are now removed from the catalog when the library opens (and by
library.forgetLocal {dryRun?, days?}), as journaled, non-undoable ops, then
compacted in the background. Files and XMP sidecars are never touched;
browsing the folder again catalogues them afresh.
Untouched = a Local file record, not deleted, not a virtual copy, whose
state still hashes to the fingerprint taken when the browse catalogued it
(any change keeps it); records from before fingerprints need every
user-editable field empty/default. Kept too: records in albums (or as
cover), stacks, sources of virtual copies, visible/selected/edited ones,
ones referenced by undo/redo, ones with a smart preview.
Per-folder last-browsed times live in the catalog (Op::SetBrowsed,
normalised with folder_key; re-stamped at most hourly). Old catalogs load
unchanged; folders without a time are stamped now, so nothing is forgotten
on the first open after the upgrade.
persist_bench (85k photos, 76.5k Local, every 7th folder recent, 1 in 50
rated): 63.8k forgotten, plan 162 ms, snapshot 313.5 MB -> 94.9 MB.
Settings -> Performance -> Local folders.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
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 <noreply@anthropic.com>
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 <noreply@anthropic.com>
An unchanged grid frame used to copy every visible id, hash them twice
(date-run and layout cache keys), clone the date runs, build a full
Vec<Cell>, rebuild the stack index, loop over every cell for the
viewport test and, in Local folder views, look up every photo to count
the browsed ones for "Add N to My Photos".
- engine: Session keeps the visible list as Arc<[PhotoId]> with a
process-unique generation that changes whenever the list is
recomputed; `visible_shared()` hands both out without copying.
- grid: date runs, layout, stack index and the Local count are cached
by that generation (GridCache; replaces the id-hashing date_runs and
grid_layout caches). The layout gains a row index (GridRow), so the
visible + prefetch rows and the date headers are found by binary
search; the sticky header too.
- GridStats counters (frames, cells visited, view/layout/stack-index
builds, last frame time) for the bench and tests.
Measured with the new `cargo run --release -p lightcraft-ui-egui
--example grid_bench` (synthetic library, 1400x900, thumbs 160, missing
files so no decoding is timed, scrolled to the middle, 120 unchanged
frames; machine at load ~200-250, so medians of the whole app frame,
before/after interleaved in the same session):
photos layout source median frame ms allocs/frame cells visited
before -> after before -> after before -> after
200 photo all 1.29 -> 1.28 4747 -> 4729 200 -> 93
200 square all 1.09 -> 1.09 4771 -> 4753 200 -> 74
10 000 photo all 1.52 -> 1.41 5425 -> 4911 10000 -> 100
10 000 photo local 1.90 -> 1.42 5475 -> 4948 10000 -> 100
10 000 square local 1.62 -> 1.07 5365 -> 4838 10000 -> 76
85 000 photo all 2.41 -> 1.29 8996 -> 4729 85000 -> 93
85 000 square all 2.29 -> 1.08 9020 -> 4753 85000 -> 74
85 000 photo local 17.6-34.0 -> 1.32 9049 -> 4766 85000 -> 93
85 000 square local 16.2-25.0 -> 1.10 9073 -> 4790 85000 -> 74
Frame cost no longer depends on the number of results; what is left
(~1 ms, ~4.7k allocs) is drawing the ~75-100 cells near the screen.
Headless snapshots of the demo grid (justified with a sticky header,
after keyboard navigation, square) are byte-identical before/after.
Tests: layout rows cover every cell in order and the binary search
matches a linear scan (square + justified, groups); a 3000-photo
headless library checks that unchanged frames build nothing and visit
< 300 cells, plus date headers, scroll-to-selection, stack badges,
keyboard navigation and a rotation re-laying the grid.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
library.import mode `move` (UI: Transfer ▸ Move, MCP import tool): reuses
Copy's destination, folder organisation (incl. custom templates) and rename
template. Each photo and its XMP sidecars (both namings) are placed without
touching the source (hard link on the same volume, else a new file that is
copied, synced and compared byte for byte; taken names get -1, -2…, never
overwritten). Sources are removed only after the catalog records are
committed and saved, and only when the destination is still intact; failed,
duplicate, unchecked and in-library files keep their sources; sources that
can't be removed are kept and reported (`kept`). Undo removes the catalog
records and leaves the moved files at the destination.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
GPX 1.0/1.1 reader in lightcraft-meta (timed track points per segment; interpolation only within a
segment, nearest point within maxGap otherwise). Engine command photo.autoTagTracklog (path or gpx
text, camera time-zone offset, maxGap, replace, dryRun; one undo step) and Photo ▸ Auto-Tag from
Tracklog… with a time-zone prompt.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
gpu (its render already runs under catch_unwind and falls back to the CPU):
- a failed readback (device lost, out of memory) logs, marks the GPU
broken and returns zeros, so crate::render discards the result and
renders on the CPU, instead of panicking;
- an unknown kernel name does the same instead of panic!;
- finish-parameter fields are looked up as Option (debug_assert on an
unknown name) instead of panic!;
- the warp sampler matches (mode, warp) like the CPU pipeline;
- Buf::raw keeps one documented expect (Some from construction to Drop).
merge:
- pyramid building (push_pull, Blender::finish), camera graph growth,
bundle adjustment, focal median and alignment use let-else / filter_map
instead of expect;
- synth::bracket_dngs / pano_views return Result (a missing demo scene
or a DNG write error is an error); lightcraft-cli synth-merge and the
engine merge tests handle it.
lightcraft-gpu, lightcraft-merge and lightcraft-cli now #![deny]
unwrap/expect/panic/unimplemented/todo/unreachable; clippy.toml as in the
steering PR.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Session::execute (the entry point of the UI, CLI, control channel and
MCP) runs each command through lightcraft_engine:💂:catch: a panic
that escapes a command becomes that command's error (EngineError::Other,
shown like any failed command) and the session, catalog and undo history
stay usable. The desktop app installs guard::install_hook, which keeps
the default stderr report and appends the panic location and message to
<temp>/lightcraft-panics.log.
The desktop app crate (already clean) now #![deny]s unwrap/expect/panic/
unimplemented/todo/unreachable; clippy.toml as in the steering PR.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- merge output_path: a bounded search for a free name instead of
`for k in 1..` + unreachable!().
- library open: Option::insert returns the new library, no expect.
- relink: a one-op batch unwraps through <[Op; 1]>::try_from, no expect.
- lightcraft-mcp (already clean) now #![deny] unwrap/expect/panic/
unimplemented/todo/unreachable; clippy.toml as in the steering PR.
crates/engine/src/export.rs keeps one expect (open PR #51 edits that
file), so lightcraft-engine doesn't get the deny attribute yet.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Rust 1.99 (stable since 2026-09-28) adds clippy::chunks_exact_to_as_chunks and widens
clippy::question_mark, which failed `cargo xtask ci` on main. Same behaviour: the
remainder that chunks_exact skipped is the remainder as_chunks returns separately.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
File now opens with Import Photos… (Cmd+Shift+I), Import from Folder… and
Import from Device (command ids file.addPhotos / file.addFolder /
file.addFromDevice unchanged; library.import's label is Import Photos).
The review dialog is titled Import Photos, names its source (a scanned
folder is not added to Local; Local -> Browse Folder... is the persistent
location), labels the transfer mode and explains Add in place vs Copy, and
its button reads Import N Photos.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
library.import organize now also takes a folder template, e.g.
{date:%Y}/{date:%Y%m%d} -> 2026/20260114 (Organize::Template). The
template's own / (or \) make the folder levels; each level is expanded
with the rename tokens and sanitized, so metadata can never add a level or
climb out of the destination; absolute paths, drive letters, ~ and . / ..
levels are refused. Dated by capture time, else the import time; a level
whose metadata is missing becomes 'unknown'. The import dialog's Folders
menu gains 'Custom template...' with a template field, tag help, live
validation and an example full destination for every mode; the dialog
stays open (with a toast) when the import is refused. The MCP import
helper passes mode/destination/organize/rename through (docs/mcp.md).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
rename::TOKENS / DATE_DIRECTIVES / TEMPLATE_NOTES are now the one list of
file-name template tags (a test checks expand_tokens knows each, and that
the rename/import command descriptions mention them all). A Tags toggle
beside the import Rename field (also Rename Photos and export File name)
lists every tag with its meaning and a live example, the {date:...}
directives and the rules (extension appended, blank keeps names, missing
date/metadata, unknown tags stay literal); clicking a tag inserts it at
the text cursor. Unknown tags are flagged under the field. New query
command photo.renameTokens returns the same help as JSON. The import
preview now uses the probe's metadata and the import time for undated
photos, as the import itself does.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Catalog Meta gets copyright_status (unknown / copyrighted / publicDomain), usage_terms and
copyright_url, read and written as XMP Rights Management fields (xmpRights:Marked, UsageTerms,
WebStatement) on import, sidecars and exports (kept by "copyright only"). photo.setMeta,
copy/paste metadata, metadata presets, a Copyright Status smart-album rule and the Info panel.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Menus are rebuilt every frame, and File → Add from Device called
devices(), which scanned every mount (D:..Z: on Windows) while holding
the cache mutex once its 2 s cache expired. devices() now returns the
last scan at once and starts at most one background rescan when it is
stale; the scan runs without the lock. A changed result calls an
on_change hook (the UI requests a repaint), so hot-plugged cards still
appear. library.devices and Add from Device without a path scan
synchronously (devices_now). Volume roots are probed for DCIM directly
instead of listing them (Windows never lists a root). No threads or
clock on wasm.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A sidecar's exif:DateTimeOriginal / photoshop:DateCreated / xmp:CreateDate
now reaches Photo.captured on import and on Read Metadata from File when
the file has no embedded capture time (embedded wins). On import the
sidecar is read before copying, so the date also files and names copies.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Missing Photos (sidebar count, the view, library.missing, Find Missing)
now shares one scope rule, cmd::missing::checked_path: non-deleted
library photos backed by a file. Local browse records, which can
outnumber the library by orders of magnitude, are never stat'ed. The
Missing view checks only the photos its query kept instead of every
catalog record.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Smart previews always lived in <library>/Smart Previews, so a large set of proxies
filled the library's drive. The folder is now a per-library setting (prefs.json):
- library.smartPreviewsLocation shows the effective folder (path, count, size,
availability) or changes it; changing it needs an explicit choice for the
previews already built (move, leave or discard), and reset returns to the
library's own folder.
- The new folder must accept writes before anything moves. A folder whose parent
is missing (an unplugged drive) is an error and is never created on the current
drive; building also refuses while a chosen folder is unavailable.
- Building, discarding, status and offline rendering all use the chosen folder.
- Settings -> Performance -> Smart previews shows the path before building, what
it holds, the existing-previews choice, Choose Folder... and Use Library Folder.
The thumbnail cache stays in the library.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A filter set from the sidebar (By Date), the search field or the filter bar
narrows the current source, but nothing said so: a folder with a hidden date
filter looked empty.
- Chips under the grid header, one per active filter, each with an X that
clears only that filter; "+N more" lists the ones that do not fit; Clear all.
- The header counts "N of M photos" (matching of the source's total) while a
filter is on; the total is cached per source/catalog revision.
- A badge on the top-bar filter icon counts the active filters, filter bar
open or not.
- An empty result says no photos match the active filters, and how many the
source holds, instead of implying the folder is empty.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
rename::expand_tokens is shared by Rename Photos, import renaming and export naming, so {seq:N},
{date:%…}, {camera}, {title}, {ext} work in exports too (a bare {seq} still pads to 3 digits there).
New shared tokens: {num} (number at the end of the name), {folder}, {lens}, {iso}, {rating},
{creator}. An export name that expands to nothing keeps the original name. The MCP export tool
passes startNumber through and documents the tokens.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Browse Folder listed, probed and imported the whole folder on the UI thread, so a
network share froze the window with no feedback. It now lists and reads on a worker
(progress window with Cancel), then adds the photos in small batches reusing the
scan's probes. Folders picked with Browse Folder… stay in the Local list after
browsing elsewhere.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A Point Curve dropdown under the curve graph shows the preset the photo's
point curves match (or Custom) and lists our own built-ins (Linear, Medium
Contrast, Strong Contrast: symmetric S-curves, values chosen here) and the
user's presets; Save Point Curve... names and stores the current four
channels, right-click deletes, Import/Export Presets... read and write
.lccurve JSON files. Presets set only the point curves (parametric curve and
Refine Saturation stay) and are saved with the library preferences.
Engine commands: curve.presets, curve.applyPreset, curve.savePreset,
curve.deletePreset, curve.importPresets, curve.exportPresets; UI commands
file.importCurvePresets / file.exportCurvePresets (file dialogs on desktop).
Fixes#14
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
New engine command curve.reset {channel?: all|point|parametric|master|red|
green|blue} (default all = the whole Tone Curve), one undo step, usable from
CLI / MCP / control channel. In the editor: double-click a channel selector
resets that channel, a Reset button under the graph resets every curve
(point curves and parametric), and right-clicking the graph offers
Reset <channel> / Reset All Curves.
Fixes#13
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Raw variants the decoder can't read yet (e.g. Nikon Huffman-compressed NEF
from a D3200/D5100) silently fell back to the camera's embedded JPEG, so
users edited a JPEG with the camera's picture style (e.g. Monochrome) baked
in without being told.
- catalog: Photo.preview_only (serde default; old catalogs load), set /
cleared by Op::SetContent with an exact inverse; Photo::develops_raw().
- engine: ProbeInfo.preview_only carries the decoder's Unsupported reason
into import, the import review (previewOnly) and catalog.query
(previewOnly); photo.inspect includes it. photo.reload and photo.relink
re-probe and clear/set it when the file decodes (or not) now.
- source_info: a preview-only raw is a rendered source (relative WB scale,
display tone curve) instead of raw:true/5500 K + filmic tone map on a
display-referred JPEG; import uses the JPEG default preset for it; the
Edit panel shows the rendered-file WB sliders.
- UI: "Preview" badge on grid thumbnails, a pill on the loupe, a notice in
Edit and Info (amber caution token, Info icon drawn in code).
- Tests: catalog serde/undo, engine import -> reopen -> reload/relink,
headless UI badge/notice widgets.
Refs #10
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Nested <dict>/<array> recursed without limit; a crafted .lmp with deep
nesting would crash the app on import. Reading stops with an error past
64 levels. Test: hostile_nesting_is_an_error_not_a_crash.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
An .lmp look is an XML property list of adjustment layers with named
sliders (newer looks: a Name.lmp/Contents/preset.lmp bundle); an
.mplumpack is a zip of looks plus PresetsInfo.plist (GroupName).
preset.import, the file dialog and drag & drop now accept both. Sliders
with a clear counterpart (exposure, contrast, highlights/shadows,
whites/blacks, WB shift, saturation/vibrance, clarity/structure,
dehaze, HSL, point curves, vignette, grain) map onto develop settings;
the rest is reported in `unmapped`. Looks are named after the look and
grouped by the pack. Tests use synthetic looks and packs only.
Fixes#9
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Brandon Thomas
storybored379
Clément L