Works through the five CodeRabbit threads on d9102e7e0a:
- A COLR graph that reuses both children of every composite was walked
exponentially often (skrifa 0.44 only limits depth and cycles on the
current path), stalling rendering and measuring. photocraft-text moves to
skrifa 0.47, already in the lockfile, which caps a walk at 16384 paints;
such a glyph falls back to its plain outline like a cycle does.
- Glyphs hidden by a transparent text colour are skipped before their graph
is walked or painted, so they can't spend the budget and drop the
visible text.
- In CMYK (and other non-RGB) documents, pixels in the text colour alone
keep its own components instead of a round trip through sRGB, as plain
glyphs do (a K-only black stays K only).
- Nested clips intersect by the smaller coverage, so a glyph clipped by its
own outline twice keeps its anti-aliased edge.
- A radial gradient's shared point of touching circles (a = b = c = 0) is
painted with the largest t instead of dividing 0 by 0.
Each fix has a test that fails without it.
Co-Authored-By: adi-1108 <adi-1108@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Playback reads the map path a loaded action file names. An action set is
shared like a document, so the path now follows the same rule as paths
stored in documents: a local absolute file. A network share, device, URL
or relative path stays a kept step with its bytes, so playing a
downloaded set never opens a remote share. Test paths are now native to
the platform, so the round trips also hold on Linux and macOS.
Co-Authored-By: yoriswind <yoriswind@yahoo.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Conflict in crates/ui-egui/src/lib.rs: main added
mod cam_no_pixels_alert_tests next to this PR's
mod eyedropper_context_menu_tests; both are kept, each with its own
#[cfg(test)].
Co-Authored-By: Hector Villarreal <hectorviov@gmail.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Main's #11972 (shield) added its own test module next to this PR's; both are
kept. Its shield test builds a Drag::New, which now also carries square and
centred: set to false (no modifier), so it compiles with this PR's pointer.
Co-Authored-By: hectorviov <hectorviov@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A warp draws each tile in row order, as one that can't fold does, and notes
whether two parts of the picture meet at a pixel; only such a tile is drawn
again in the fold's order, with its layers composited, and the order is
worked out the first time a tile needs it. Without a fold the order makes no
difference, so such a warp costs what it did before folds stacked.
24 MP layer, release, best of three: main 0.22 s without a fold, 0.23 s with
one; now 0.20 s and 0.26 s (was 0.24 s and 0.29 s).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01EZbgszHyuEuWAzxaamteuU
While editing with any of the four type tools, Cmd+. (Ctrl+. off the
Mac) now does what the options bar's Cancel button does, as in the
reference application: the session ends and the document, the
selection and the history are as they were before it.
Shift+Cmd+. stays Increase Type Size. Outside a type session, while an
input method is composing and during a Cmd transform drag (not
measured) the key does what it did before: nothing.
Timing a 24 MP Warp showed folds cost five times main's time (1.23 s
against 0.22 s): a tile where the dragged part lies over the rest samples
two distant parts of the source, and its one window spanned everything
between them (230 MP read per render against 31 MP). A tile's samples are
now grouped into connected regions of the source on a coarse grid, each
read as a window of its own; a tile whose samples lie close together stays
one window, without the grid. A fold's layers are kept in one linked list
per tile instead of a list per folded pixel, and the cells are sorted in
parallel.
24 MP layer, release, best of three: main 0.22 s without a fold and
0.23 s with one; before 0.28 s and 1.23 s; now 0.24 s and 0.29 s.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01EZbgszHyuEuWAzxaamteuU
A corner moved one pixel could take a quad across the parallel-sides limit and the output from 602 x 400 to 719 x 289. The estimate now has to survive one pixel of error on every corner, with its proportions moving by no more than 10 %. Also test the commit path with W and H empty, and a known rectangle seen by a camera model.
The earlier people.toml merge resolution inserted the new
[people.Kaedo17] header between [people.davidvhk]'s real_name and its
sync_github_name = true, so TOML parsed that setting as Kaedo17's and
davidvhk silently lost theirs. Restore davidvhk's setting and keep
Kaedo17 as its own block with its own sync_github_name = true
(FredApps, #9841).
Parsing base 9394960e4 and this tree side by side: the only difference
is the added [people.Kaedo17] entry; every existing entry is unchanged.
Related to #5310. Display labels only; command ids (filter.cameraRaw, edit.preferences.rawDefaults), preference keys, serialized settings, PSD/action/XMP Adobe Camera Raw identifiers and .atn step names are unchanged.
- Camera Raw / Camera Raw Filter / Camera Raw Defaults -> RAW Develop / RAW Develop Filter / RAW Defaults in menus, preferences, filter names, diagnostics and all 34 catalogs.
- Saved Photoshop actions keep the "Camera Raw Filter" step name, so .atn files round-trip byte for byte (test extended).
- "Camera Raw" stays a command-search keyword.
- Removes dormant Adobe/Photoshop-branded catalog rows; display-name guard tests.
Right-clicking a pixel layer's thumbnail in the Layers panel opens
Photoshop's thumbnail menu instead of the layer menu: Select Pixels, which
loads the layer's transparency as the selection as a Cmd/Ctrl-click on the
thumbnail does, then the thumbnail sizes and Clip Thumbnails to Layer /
Document Bounds (shared with the empty-area menu, #11404). The Background,
with no transparency to load, frames, and type, shape, smart object, fill,
adjustment and group thumbnails keep the layer menu with their own items.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Protect Foreground Color and Sampling Background Swatch converted the
foreground and background from their sRGB stand-ins, so a colour picked
as CMYK inks (100/0/0/0) came back as other inks and pixels painted with
exactly those inks were erased or missed. BgEraseSettings now carries the
picked inks, as BrushSettings::color_cmyk does, and a CMYK target compares
against them while the colour is still that sRGB colour. RGB, Gray and Lab
documents are unchanged.
The Shift-click override only reached the Brush, Pencil, Eraser and Mixer
Brush. The retouching tools take their brush from the engine
(`retouch_cmds::stroke_brush`, or `resolve_brush` for the Background Eraser
and Color Replacement), so with Catch-up on Stroke End off their lines
stopped short of the click.
- `stroke_brush` takes `smoothing` (0..100, the options bar's Smoothing).
- `canvas::unsmoothed_line` sends `smoothing: 0` for a Shift-click line in
every retouching tool's live preview and in both commits, so a commit
still takes the pixels its preview drew.
cargo fmt --all -- --check on the merge commit flags only the
glide_scroll_all assignment added by the conflict resolution: the merge
produced it as one long line, and rustfmt wants the call on its own line.
No behaviour change (cloutier, #9841).
The Set of the document's measurementScale property (T: pixelLength long,
logicalLength double, logicalUnits text) loads as
image.analysis.setMeasurementScale with the custom preset and saves back
with its recorded descriptor. Other forms stay kept steps; a native step
that reads, resets or partly sets the scale refuses export.