Review follow-ups:
- Check every PointerMoved / PointerGone of the frame, not only where the
pointer ended: a slow frame merged leaving the row and coming back, so
the slider stayed armed (new test for both events).
- One global armed slider (`wheel_slider_key`) instead of a bool per
slider: unarmed sliders no longer write context data every frame, and
arming another slider disarms the previous one. The wheel remainder is
cleared on arming and disarming, so the second focus check is gone.
- The sideways branch uses `pointer_in_row` too.
- Document that Tab focus arms the arrows only; move the whats-new entry
to 2026-10-21; update the KEY-SLIDER-NUDGE row, header and history; fix
a test comment that described the old remainder carry-over.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The PR branch dropped it along with the PREVIEW_LIMITS import; nothing
here changes the default preview limit.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Folder templates (and rename / export names) can now give month folders by
name: {date:%Y}/{date:%B} → 2026/January, {date:%b} → Jan. The names are
English in every interface language so a library's folders don't change
with the language; a month that doesn't parse gives nothing (a folder
level becomes "unknown").
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
An ASCII path and rule text are compared byte by byte, folding case and
backslashes, with no allocation; other text takes the old code unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The web release now builds with craft-fonts too (web-craft-fonts), as VectorCraft's
does, so the published web app ships the Chinese and Japanese faces docs/releasing.md
promises.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The random-edit test now runs five seeds instead of one. A grid-mode change
re-reading every photo, and the group names growing unpruned, are now said in
the comments (the "a day or two at most" claim was wrong for capture-time edits).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every catalog change gave the grid a new list of photos, and it read every
photo of it again: each one's date group (a lookup, a date parse and a
string) and its cell's shape. At 180,000 photos that was a 100 ms frame for
every batch an import added.
- catalog: Catalog::date_runs is split into date_group (one photo's own
group) and date_runs_of (the stack rule and the runs), which the grid
uses too.
- ui: GridCache keeps what it read of each photo of the list, position by
position. A new list that differs by the photos the catalog says an edit
touched is read again only for those; anything else (another view,
order or grouping, a stack) is read whole. The square grid reads no
shapes. A 16-photo batch at 181,749 photos: 101 ms -> 7 ms.
Co-Authored-By: Claude <noreply@anthropic.com>
The cancelled result is never used, but scan_files still checked each listed file's cached probe against the file.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The row read "1 folders, 1 files"; it now reads "folders: 1, files: 1" (all 13 catalogs).
`listed` is stored with Release and loaded with Acquire, so a reader that sees the listing over also sees what was stored before it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The scan lists the whole folder tree before it checks a photo. On a large
folder on a network share that takes minutes, and the scan's row showed
"0 of 64085" with an empty bar all that while: it looked frozen. ✕ did
not stop the listing either.
- engine: SyncProgress counts the folders and files the walk listed and
says when the listing is over; a folder note (walk::NoteFolder) now
answers whether to go on, and a cancelled scan answers no.
- ui: while the folder is listed, the row has an indeterminate bar and
reads "Listing folders: N folders, M files · folder" (all 13
catalogs); the dialog says the same; then the photos checked are
counted as before.
Co-Authored-By: Claude <noreply@anthropic.com>
When the scan finds nothing to import, relink, remove or read, the dialog
showed four greyed choices (two still ticked) and a Synchronize button
that started an empty job with an activity row. It now shows the folder,
"The library is up to date with this folder." and any Skipped lines, with
one Close and no Cancel (`only_shows`). `sync::confirm` answers
`{"upToDate": true}` and starts nothing, so `ui.dialog.confirm` and agents
get the same result.
Co-Authored-By: Claude <noreply@anthropic.com>
egui clears `press_origin` on the release, so a slow frame that swallows
press, move and release into one batch reports `primary_pressed` with no
origin: the gesture fell back to the release position and a real drag
wrote nothing (its length check saw a zero-length stub). The gate now
reads the primary press back from the frame's own events before that
fallback — the press decides what it means, wherever the batch ended.
`contains_pointer` reads where the pointer is NOW; a slow frame can
deliver the press on a menu and the move off it in the same pass, and
the gate would then pass with the saved press origin under the menu.
`layer_id_at(press_origin) == resp.layer_id` asks which layer owns the
press itself: the menu's area (or any modal) wins there and the guide
under it stays untouched. The navigator check stays for its own widget.
`primary_pressed` plus `resp.rect.contains` reads the raw pointer, so
anything floating above the photo — the context menu, a dropdown, a
dialog — also started a guide gesture under it. The gate now asks
`resp.contains_pointer()`, which the hit test keeps honest about
covering widgets; the navigator check stays. Two tests press on a menu
over a line and over an endpoint: nothing is selected, moved or drawn.
- a secondary/middle press or one on the navigator panned the photo but
also selected or drew a guide: the gestures take the primary button and
skip the navigator's rect
- egui forgets the press origin by the release, so the endpoint drag
compared against the guide's old position: a press that landed within
the grab radius and never moved wrote a no-op guide; the gesture now
carries the press point itself
- a selection left behind by a Delete past the last guide clears with the
photo/tool check in crop_overlay
Split out of #1887 (part 2 of 3, stacked on part 1). The modules land on their
own; the HDR merge starts using them in part 3.
- refine.rs: an alignment found on <= 1200 px versions refined at full size.
One 48 px patch at the most textured spot (smaller structure-tensor
eigenvalue) of each of ~24 cells along the long edge, matched by masked
normalised cross-correlation over the pixels usable in both frames with a
sub-pixel peak (ambiguous and textureless patches dropped), fitted by RANSAC
at 0.75 px with the simplest of translation / similarity / homography. Also
recovers a frame with no alignment, searching +-16 px.
- flow.rs: what one homography leaves (parallax, distortion), measured on ~96
tiles along the long edge within +-5 px, outliers against their neighbours'
median dropped, gaps filled, the field smoothed; flow::warp resamples once
through the homography plus the local shift (Lanczos-3). align::warp is
flow::warp with no shift.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A press starts what it means right away from the pointer's press point: on an endpoint (within 7 px) it drags that endpoint, committing one geometry.guides edit on release (the guide must stay at least 0.02 long; a press without a real move only selects, so a click never leaves a no-op undo entry); on a line (within 6 px) it selects the guide (a press on the selected one deselects); anywhere else it starts a new guide from the press point itself instead of where egui's drag threshold was crossed, leaving nothing selected. The endpoint being dragged is drawn bright while the pointer moves. tests_guided_guides 8/8.
A click near a guide selects it (white, thicker, larger endpoints); drawing a new guide or clicking away clears the selection, and so does leaving the photo or the tool. Delete / Backspace with the tool open removes only the selected guide (geometry.guides minus its index, one undo step); with none selected it removes nothing — the photo itself is never deleted while Guided Upright is open. selected_guide is UI-only (serde skip). Coverage: tests_guided_guides 5/5.
with_fs() now sets media.reduced_loader, which thumbnail jobs prefer over file_loader, so the render
fixture's stub file_loader was bypassed: the job read the nonexistent audit.jpg and
resizing_the_disk_cache_keeps_textures_and_skips_rebuilds got no texture (on every OS, not a flake).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Split out of #1887 (part 1 of 3). Every warp in the merge crate (HDR alignment,
panorama projection, Boundary Warp) sampled bilinearly: at a half-pixel offset
that is a [½ ½] filter, which keeps cos²(πf) of a pattern's fine detail (a third
at 0.3 cycles per pixel).
- lightcraft_raster::sample: Rgb32f::sample_lanczos3, a 6 × 6 Lanczos-3 point
sampler with tabulated taps, each sample held to the range of the 4 × 4 pixels
around it so hard edges (clipped highlights, stars) don't ring.
- align::warp, the panorama projection and Boundary Warp use it.
- align.rs: the merge luminance (`luma`) and the simplest-model choice among
RANSAC fits (`choose_model`) are shared functions, for the full-size
refinement that builds on them.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Pure functions on a RuleSet (judge, move_node, key_dest and their path helpers)
with unit tests; nothing calls them yet. The smart album editor's reordering
(a separate change) is built on them.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Japanese photo printing and New Year cards cannot pick their paper: the Print
page setup only offers the ISO and US sizes. Add the customary papers at their
exact metric sizes, after the existing entries (whose sizes and order are
unchanged): L判 89×127 mm, 2L判 127×178, KG判 102×152, はがき 100×148,
往復はがき 148×200, 六切 203×254, 四切 254×305 and 名刺 55×91 (portrait).
Names resolve case-, space- and punctuation-insensitively, and take the
Japanese customary names as aliases (L判, Lサイズ, L-size, 2L判, KG判,
はがき/ハガキ/葉書/ポストカード, 往復はがき, 六切/6切, 四切/4切, 名刺). The
Print dialog and `print.set`/`print.get` use the same names, the labels show
the customary name with its metric size, and the PDF page size and MediaBox
follow the paper.
Tests: the layout test pins every new size to 0.1 pt, the aliases, and the
unchanged ISO/US sizes and unknown-name rejection; the Print view test checks
the labels; and an engine test exports a PDF on はがき (283.46×419.53 pt) and
L判 (252.28×359.99 pt) at 254 ppi (ten pixels per millimetre, so the MediaBox
is exact) and reads the actual page size, plus the customary name through
print.set.
Docs: ROADMAP progress log + revision history and header, docs/gaps.md
(Gap 15 revision), docs/parity-checklist.md (LRC-PRINT-LAYOUT row), dated
2026-10-22; existing revision rows keep their dates.
crates/ui-egui/locales/{cs,it,pl}-formats.json: kept the translations and added the track's new keys.
docs/parity-checklist.md: newer header line; revision rows from both sides.
docs/whats-new.md: both tracks' 2026-10-15 entries.
Held for the owner: #995 and #999 overlap, and #995's author offered on 10-12 to
close#995 once his commit was in #999. docs/parity-checklist.md: header keeps the
#1485 note and drops the #995 part.
