Commit Graph
11425 Commits
Author SHA1 Message Date
David MannandClaude Opus 5.5 33e09e56fe Merge PR #3567: History panel: New Snapshot dialog and the first snapshot (#2966)
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-25 13:37:08 +01:00
David MannandClaude Opus 5.5 8c15e55cbf Integration: merge main, resolve conflicts (#3567)
panels.rs: the PR's snapshot rows next to main's History Brush source
and scroll-to-current; ui-parity.md: main's status line.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-25 13:36:40 +01:00
David MannandClaude Opus 5.5 b53f2dcac7 Automation: refuse every Windows device name in request paths (#9424)
The automation device-name check was narrower than cap-std's own
Windows check (COM0/LPT0, COM¹-³/LPT¹-³, spaces before the extension).
write() renames through a Win32 path and skips cap-std's check, so a
file such as COM¹.png or "NUL .png" could be written but not read
back, and export_sequence left an empty folder behind.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 13:26:53 +01:00
096ad5b194 Integration: merge main, resolve conflicts (#7203)
Main moved the shape options bar and paint swatch out of vector_ui.rs into
shape_options.rs and shape_paint_ui.rs (#5452). Carry the PR's gradient paint
over: PaintChange::Gradient, paint_param and the swatch's Gradient tab live in
shape_paint_ui.rs, the tool gradients and set_tool_paint in shape_options.rs,
and shape_gradient_ui.rs is its own module. Both sides' new tests are kept.

Co-authored-by: Denis <179751626+deniskhud@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 13:26:26 +01:00
highperfocused 31a00fcde4 Merge upstream/main into feat/color-hud 2026-10-25 13:25:48 +01:00
a222c2aee4 Integration: Calibration keeps the Camera Raw preview on the CPU develop (#5128, #6923)
#6923 develops the Camera Raw dialog's preview on the GPU whenever no spatial stage is set,
from `per_pixel`'s stages. #5128 adds Calibration (primaries and Shadows tint) to the CPU
develop between white balance and Dehaze, which those stages don't run, so with both merged
a Calibration-only change previewed unchanged on the GPU while Open applied it. `per_pixel`
now returns None when Calibration is set, so the dialog develops on the CPU then. The
white and black points stay as they were, so the dialog's points cache is still right once
Calibration goes back to neutral. The hardware parity doc and BUG-204-3's note now say so.

Co-authored-by: Fabio Lissi <3223836+FabioLissi@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 13:12:57 +01:00
David Mann 7eec69f8f5 Merge PR #5128: Camera Raw: Calibration panel, Auto, raw sidecars and live Camera Raw Defaults (FILE-215-11)
# Conflicts:
#	crates/engine/src/prefs.rs
#	docs/gaps.md
#	docs/scorecard.md
2026-10-25 13:07:14 +01:00
fab7d9739e Integration: keep foreign raw sidecars, develop before the workflow conversion (#5128)
Open wrote PhotoCraft's crs: subset over any existing sidecar, on by
default: a Lightroom or Bridge IMG_0001.xmp lost its crop, masks, rating
and keywords, even when the raw was only opened with Open in Camera Raw
off (FredApps' review). Now a sidecar is only written when the settings
changed, never over one whose x:xmptk isn't PhotoCraft's (a notice says
it was kept, camera_raw_xmp::is_photocraft_xmp), and never for a control
or MCP request, which has no authority to write beside the user's files.
A write error goes to a notice instead of a status line the develop
overwrote at once.

The Camera Raw Defaults' colour space, 8-bit and smart-object steps ran
before the Camera Raw step, so with the default Adobe RGB every raw opened
with the dialog's settings applied to other pixels than it previewed. They
now follow the step (also when a background re-develop lands), as
Camera Raw converts to the workflow space last. Main already sets the raw
resolution at import (#7811), so the PR's Image Size step is dropped.

Two raws opened back to back with raw-stage edits: the second one's
background re-develop replaced the first one's pending entry, so the first
never got its Camera Raw step, conversions or unmodified mark. Pending
re-develops are now a list keyed by job. Sidecars are read only from
regular files of at most MAX_XMP_BYTES (a pipe or a huge file named
IMG.xmp no longer blocks or fills memory), and a conversion that fails
after the Camera Raw step is reported instead of leaving the dialog open
for a second Open that would apply the step twice.

Tests: a_foreign_sidecar_is_never_overwritten,
a_rating_only_sidecar_is_left_alone,
opening_without_the_dialog_does_not_rewrite_the_sidecar,
a_control_request_never_writes_the_sidecar,
a_sidecar_write_error_stays_reported,
the_camera_raw_step_runs_before_the_workflow_conversion (inline and
background), two_raws_opened_back_to_back_both_finish_their_open,
sidecars_are_read_from_small_regular_files_only,
is_photocraft_xmp_reads_the_toolkit_marker; the existing
sidecar round trip now uses a PhotoCraft-written fixture.

Co-authored-by: Fabio Lissi <3223836+FabioLissi@users.noreply.github.com>
Co-authored-by: FredApps <153173854+FredApps@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 13:03:14 +01:00
David MannandClaude Opus 5.5 2415c6979f Integration: merge main, resolve conflicts (#3627)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 13:00:08 +01:00
David MannandClaude Opus 5.5 e1b9c30294 XML: check every DOCTYPE for an internal subset, not only the prolog's
A quoted ?> in the XML declaration hid the DOCTYPE from the prolog walk.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 12:59:31 +01:00
andeecollard 5c30699087 Preserve surviving Cutout islands through low-fidelity rendering 2026-10-25 11:58:33 +00:00
andeecollard 864024609b Fix Cutout island collapse at low edge fidelity 2026-10-25 11:57:48 +00:00
David MannandClaude Opus 5.5 1c595a34d9 Address CodeRabbit review: filter cost check stops after a visit budget
Patterns that fill and stroke with the next one double the elements the
check walks at every level (and resvg's render with it): 24 levels took
about 100 s. Past 10 million visits the drawing now counts as too costly.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 12:53:26 +01:00
David Mann aee5dac19f Merge PR #8658: fix(xtask): derive scorecard corpus counts from pinned manifests
# Conflicts:
#	docs/scorecard.md
2026-10-25 12:49:14 +01:00
David MannandClaude Opus 5.5 027bfb30ed XML: deeply nested ORA, KRA, Camera Raw XMP, .kys and layer XMP are an error, not a crash (#9419)
roxmltree recurses once per element level, so a few thousand nested
tags overflow the thread stack, an abort no guard can catch. SVG checked
the nesting before parsing; the other five roxmltree callers now go
through photocraft_io::xml::parse, which applies the same check (512
levels) and, where a DTD is allowed (KRA), refuses an internal subset
whose entities could expand into nesting the check never sees.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 12:49:05 +01:00
Khawar Zamman Wani 280f7efc0f Save for Web: WebP keeps the photo's EXIF when Metadata is All
The WebP branch built its image from meta(), which carries only the resolution and XMP, and
embed_metadata followed the XMP alone, so an EXIF-only photo lost its camera data in both lossy
and lossless modes. The filtered EXIF now goes onto the image and enables the metadata when
either it or the XMP is present. Regression: an EXIF-only photo exported as WebP (both modes)
keeps its EXIF with All and drops it for every other choice.
2026-10-25 17:17:47 +05:30
David MannandClaude Opus 5.5 a60b72483e Integration: merge main, resolve conflicts (#8658)
docs/file-format-parity.md: main's PSD row with the round-trip count
258/258. docs/scorecard.md: regenerated with `cargo xtask scorecard`.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 12:46:18 +01:00
9a345bb6a8 Integration: the Photoshop corpus has 258 files everywhere (#8658)
The PR fixes the stale Photoshop corpus count (256 -> 258, the pinned
manifest xtask/photoshop-corpus.sha256 lists 258 PSDs) in
docs/development.md and docs/scorecard.md. The same stale number was
left in the io corpus test comment, docs/file-format-parity.md,
docs/roadmap.md ("258/256" round-trip floor) and ATTRIBUTION.md.

Co-authored-by: Simone <185146821+Lucenx9@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 12:42:43 +01:00
David Mann 75e2ce844c Merge PR #8175: PSD: type on a path opens on its path and saves back into the file's Txt2 (#7962) 2026-10-25 12:40:11 +01:00
highperfocused 23f7e2039f fix(ui): crop box Undo keeps the reset frame's no-op state (#6561)
The box's own Undo/Redo steps did not carry reset_frame, so Reset, then a
gesture, then Undo restored the full-canvas frame with the reset state
dropped: Commit ran image.crop again, trimming layer pixels outside the
canvas with Delete Cropped Pixels on and recording a history step. Keep
the flag in CropStep and restore it in apply_step, so an undone reset
frame commits as the no-op it was before the edit.

Regression test: reset_edit_undo_then_commit_crops_nothing.
2026-10-25 12:36:17 +01:00
highperfocused e47346b09d Merge upstream/main into t3code/fix-low-hanging-issue-10
Conflicts:
- crates/ui-egui/src/canvas_tool_menu.rs: keep both the upstream type-layer
  menu field (#7324) and this PR's crop menu fields (#6561) in the struct
  literals and the auto-close condition.
- docs/context-menu-parity.md, docs/scorecard.md, scorecard/ui.toml: keep
  both sides' status-line, revision-history and UI-217-8 note entries.

Semantic fix from the merge: crop_size::resamples called target() with the
old two-argument signature; pass the Point Size preference like the other
callers.
2026-10-25 12:36:12 +01:00
d396a80272 Integration: one Txt2 frame per layer on save (#8175)
Two layers could update the same Txt2 frame, the last one winning. An
edited layer whose object can't be refitted (a run array whose length
is no count) was rebuilt from a template naming frame <slot number>,
which can be the frame another object draws in; and a new layer whose
slot number is a path layer's frame number named that frame too. A
point layer then turned the path layer's frame into a point frame, and
the path text reopened as plain text.

A rebuilt object now keeps the frame its old object was drawn in (when
the frame list has it), and a new or rebuilt object whose slot number
is another object's frame gets a free frame past the list (padded like
the others). A frame the file already had two objects drawn in is left
as the file has it. Tests (each fails without its part):
an_object_rebuilt_from_a_template_keeps_its_own_frame,
a_new_object_gets_a_frame_no_other_object_is_drawn_in,
a_rebuilt_object_never_names_a_missing_frame,
a_frame_two_objects_name_is_kept,
a_kept_object_naming_a_far_frame_adds_no_frames (the list only grows
for frames handed out here); an_area_layer_gets_a_box_frame
covers the merge with main's area type.

Answers CodeRabbit's review thread at crates/text/src/psd.rs:1242.

Co-authored-by: Michael Higgins <55243596+Higgs32584@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 12:35:07 +01:00
David MannandClaude Opus 5.5 8250046f99 SVG: filter cost counts masks, patterns and embedded drawings at their drawn scale
Count only the reads beyond an ordinary per-pixel allowance, so ordinary
filter chains still open on the largest canvas.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 12:21:30 +01:00
highperfocused 788c8956c0 Color Lookup: honour the .3dl Mesh header's cube edge and output depth (#9078, #9073)
The Lustre `3DMESH` / `Mesh <inBits> <outBits>` header declares the cube edge
(2^inBits + 1) and the output code range (up to 2^outBits - 1); the parser
skipped every text line, so it inferred both from the data. That rejected a
3-squared cube whose three-value shaper row counted as a 28th node, and
divided 10-bit codes under a 12-bit declaration by 1023 instead of 4095.
Photoshop 27.11 accepts and renders both (measured by the reporters).
2026-10-25 12:17:57 +01:00
David MannandClaude Opus 5.5 cfb5812fb9 Integration: merge main (#8175)
Resolves the conflicts with main's area type (TextShape::Area): EngineData keeps main's area arm beside the path comment, and a Txt2 frame of an area layer becomes a text box over the path's bounds, as its EngineData describes it.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 12:14:17 +01:00
David MannandClaude Opus 5.5 b974aca9dd SVG: a drawing that could not be rendered isn't also called empty (#9404)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 12:10:39 +01:00
David Mann 6d0ab91536 Merge PR #4097: Integrations: wire Allow Agent Control and Control Port (#204)
# Conflicts:
#	docs/scorecard.md
2026-10-25 12:10:21 +01:00
David MannandClaude Opus 5.5 7d15a6b7eb Integration: blend mode keys leave the layer alone with Art History Brush (#1610)
#1610 adds the Art History Brush and #5478 the blend mode keys; neither knew the
other. With Art History Brush active, Shift+Alt+M or Shift+= fell through to
Target::Layers and set every selected layer's blend mode. Photoshop's Art
History Brush has a Mode of its own, so like the other painting tools without a
Mode control here yet, the keys now do nothing.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 12:07:35 +01:00
Federico Olivares 971d6a3a1e Write the over-budget pattern warning with writeln! 2026-10-25 08:06:47 -03:00
David MannandClaude Opus 5.5 a617ba83fe SVG: a drawing the renderer panics on is an error, not a crash (#9404)
usvg, resvg and tiny-skia panic on some malformed drawings (an feOffset
past the pixmap's size, an embedded SVG image of near-zero width). Parse
and every render now run under catch_unwind, as the HEIF and JPEG XL
decoders do: a parse panic is IoError::Svg, a panicking element is dropped
with a warning and the rest opens, a smart-object render is an error, and
an icon-theme mask is None.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 12:05:11 +01:00
David MannandClaude Opus 5.5 c1c36251cb SVG: refuse filters too costly to render (#9409)
resvg's feMorphology reads the whole radius window per pixel, feConvolveMatrix
the whole kernel, and feTurbulence runs every octave (usvg doesn't clamp
numOctaves), and none of it can be interrupted: a small file kept opening or a
smart-object render busy for minutes. Estimate the filter work before rendering
(masks, clip paths, patterns and embedded drawings included, at the render
scale) and refuse the drawing past about ten seconds of it.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 12:03:41 +01:00
David Mann f85ff03a26 Merge PR #1610: Art History Brush in the Y-group flyout
# Conflicts:
#	crates/ui-egui/src/retouch_ui.rs
#	docs/brush-parity.md
2026-10-25 12:03:30 +01:00
8208034bef Integration: Integrations prefs read leniently, dialog test and restart note (#4097)
- prefs_ui `unimplemented_preferences_are_hidden_from_the_dialog` still asserted the
  Integrations section is hidden, so test (ubuntu/windows) failed (FredApps' review).
  It now asserts both rows are visible; scratchDisks stays hidden.
- `integrations()` parsed the whole preferences file strictly, so one wrong-typed value
  anywhere (`{"interface":{"enableAnimations":"nope"},"integrations":{"allowAgentControl":false}}`)
  reset Allow Agent Control to on while the dialog (lenient load, #6611) showed it off.
  The section is now read on its own, and field by field when it doesn't read as a whole
  (a list like `[false]` reads as a whole, as in the app); a test compares it with the
  app's own load.
- Both rows are read only at launch: they say "Applies at next launch." like System Title Bar.
- Removed the stale "not by these yet" comment in HIDDEN_UNTIL_IMPLEMENTED.

Co-authored-by: Erica Franco <20589100+Erica4fr@users.noreply.github.com>
Co-authored-by: FredApps <153173854+FredApps@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 12:02:55 +01:00
David MannandClaude Opus 5.5 028d400454 Integration: merge main, resolve conflicts (#5128)
prefs.rs keeps main's demosaic default and the PR's write_xmp_sidecars, and
unhides the Camera Raw Defaults the PR makes live; the raw-open RawOpen built
without the dialog gains main's developed_with/demosaic; i18n and docs as
unions; docs/scorecard.md is regenerated in the fix-up.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 12:02:10 +01:00
vangel76andClaude Sonnet 5.5 742b068052 Affinity reader: keep curve records in one buffer and index objects without a hash map
A fixed-size record array (0x2c, the curve nodes) allocated a Vec per record.
It is now one Value::Records buffer; each record still counts as one value
against MAX_VALUES. Curves read the nodes straight from the buffer, and the
other readers of array fields (slot counts, gradient positions, column widths)
handle the new variant the same way as the array of byte values it replaces.

Every object also built a SipHash map of its tags. Objects with up to 12 fields
are now scanned, larger ones use a sorted (tag, first index) table with a binary
search, so lookups stay logarithmic and keep first-match semantics.

read() also drops the inflated doc.dat once the stream is parsed instead of
holding it while the model is built.

Part of #9413

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 11:55:21 +01:00
vangel76andClaude Sonnet 5.5 44fd2ea784 Affinity reader: decompress the tiles of a plane on worker threads
After the conversion fix, zstd decoding is about 80 percent of raster time and
every tile was decoded one after the other. A plane now walks its tile grid
first (flat cells are filled on the way), then extracts all tiles it needs with
the new Archive::read_many, which decodes on up to 8 std::thread::scope threads
(entries are taken from a shared counter), and copies them in place.

Limits are unchanged: read_many charges every entry in order before decoding,
exactly as that many read() calls, and returns the first error in that order
(a failing tile before a refused or malformed later one wins). Fewer than four
tiles, wasm32, or a failed thread start fall back to decoding on the calling
thread. Output is byte-identical.

Part of #9412

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 11:51:52 +01:00
vangel76andClaude Sonnet 5.5 ace68fbb14 Affinity reader: convert raster tiles with tables instead of a per-pixel Vec<f64>
Each pixel of a decoded layer was widened to a freshly allocated Vec<f64>,
divided, clamped and rounded through libm. The channels are now interleaved
directly: 8-bit RGBA and grey are plain copies (x / 255 * 255 rounds back to x),
16-bit goes through a 64 Ki table and CMYK through a table over (c, k), both
built from the same f64 expressions, so the bytes are identical. Tile copies
into a plane use one slice copy or fill per row instead of a closure per byte.

The scratch charged to the pixel budget is unchanged. Tests keep the old
conversion and compare it exhaustively (every byte, every 16-bit value, every
CMYK colour/black pair) and the old byte copy against the row copy for many crops.

Part of #9412

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-25 11:45:21 +01:00
Hector VillarrealandClaude Opus 5.5 f241bba1ad Saves keep the creation time under a deny-write ACL and through automation
CodeRabbit (discussion_r4370498176, two findings):
- copy_metadata copied the ACL before the creation time; an ACL that
  denies its owner write access, copied onto the temporary file first,
  made the creation-time step fail to open it. The time goes first now.
- AuthorizedWorkspace::write (the automation save path) copied only the
  extended attributes, so an automation save reset Finder's Created date.
  It now also copies the creation time through the same capability-checked
  handles, with the new descriptor-based meta::copy_created_file.

Tests: keeps_the_creation_time_under_a_deny_write_acl,
write_keeps_the_replaced_files_creation_time.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 04:43:52 -06:00
Matt Van Horn cf1530a5ae Merge commit 'b0622d20' into fix/art-history-brush
# Conflicts:
#	crates/ui-egui/src/i18n/cs.tsv
#	crates/ui-egui/src/i18n/de.tsv
#	crates/ui-egui/src/i18n/el.tsv
#	crates/ui-egui/src/i18n/es.tsv
#	crates/ui-egui/src/i18n/fr.tsv
#	crates/ui-egui/src/i18n/hu.tsv
#	crates/ui-egui/src/i18n/id.tsv
#	crates/ui-egui/src/i18n/it.tsv
#	crates/ui-egui/src/i18n/ja.tsv
#	crates/ui-egui/src/i18n/ko.tsv
#	crates/ui-egui/src/i18n/nl.tsv
#	crates/ui-egui/src/i18n/pl.tsv
#	crates/ui-egui/src/i18n/pt-br.tsv
#	crates/ui-egui/src/i18n/ru.tsv
#	crates/ui-egui/src/i18n/sr.tsv
#	crates/ui-egui/src/i18n/tr.tsv
#	crates/ui-egui/src/i18n/uk.tsv
#	crates/ui-egui/src/i18n/vi.tsv
#	crates/ui-egui/src/i18n/zh-hans.tsv
#	crates/ui-egui/src/i18n/zh-hant.tsv
#	docs/scorecard.md
2026-10-25 03:25:22 -07:00
David MannandClaude Opus 5.5 b05e02c9b3 SVG: end the DTD's element, attribute and notation declarations as the parser does
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 11:20:04 +01:00
Hector VillarrealandClaude Opus 5.5 89d7d79797 fix(format): a save on macOS keeps the file's creation time (#9388)
Saving writes a temporary file and renames it over the original, so on
macOS the saved file was born at the time of the save and Finder's
"Created" date moved forward. Photoshop keeps it.

`copy_metadata` now copies the replaced file's creation time onto the
temporary file before the rename, with std's `FileTimesExt::set_created`
(stable since Rust 1.75). Best effort, like the other metadata steps.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 04:16:44 -06:00
David Mann 8fd157afef Integration: merge main, resolve conflicts (#4097)
docs/scorecard.md regenerated with cargo xtask scorecard.
2026-10-25 11:15:07 +01:00
David MannandClaude Opus 5.5 e17d6d5f4d SVG: the embedded SVG shares the drawing's nesting budget
usvg parses the embedded SVG on top of the drawing's own recursion, so
511 + 511 groups, each fine alone, still overflowed a 1 MB stack. The
embedded SVG now gets MAX_XML_DEPTH minus the drawing's depth.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 11:14:29 +01:00
Matt Van Horn 9db1c705e4 Art History Brush: clip the source window and fix the tolerance default
Clip each dab's source window to the dab's bounds plus one pixel for
bilinear sampling, so a large brush no longer allocates and copies the
whole tile-plus-halo window for every tile it overlaps. Pixels are
unchanged; a new test stamps with an oversized halo and compares.

Tolerance is a percentage of the channel range, and the default is now
0 (paint everywhere), as in Photoshop. At the old default of 50 a dab
was skipped unless the history state differed by half the channel
range on average, so most strokes painted nothing. The difference
average now leaves alpha out, so an unchanged alpha no longer dilutes
a colour change. A stroke whose every dab falls under the tolerance
returns with no damage and records no history step.
2026-10-25 03:11:53 -07:00
ce6b831c03 Integration: Art History Brush tolerance, source window, History source (#1610)
- Tolerance defaulted to 50 and averaged alpha into the difference, so an
  ordinary edit (e.g. #c03030 -> #d04848) was never painted back, and a stroke
  that wrote nothing still left an empty History step. Photoshop's default is
  0; the difference is now the larger of the mean colour change and the
  alpha change (the Eraser keeps colour and lowers alpha), and a stroke that
  writes nothing is an error instead of an empty step.
- Each tile copied a source window of tile +- halo per dab: a leftward stroke
  or a curl across x = 64 read empty pixels there (seam), and a size-1000 click
  copied ~5e9 floats (11.9 s). The rotated read stays inside the dab, so it now
  samples the source directly; bilinear reads at pixel centres, so an
  unrotated dab copies the history pixel instead of a 2x2 blend.
- The brush used its own history lookup (state 0 by default): it now takes
  the History Brush's source (history.setSource, snapshots, refusal after a
  canvas size change).
- TOOL-213-20 is partial (8 of 10 styles, Area scales spacing); docs follow.

Co-authored-by: Matt Van Horn <455140+mvanhorn@users.noreply.github.com>
Co-authored-by: FredApps <153173854+FredApps@users.noreply.github.com>
Co-authored-by: coderabbitai[bot] <136622811+coderabbitai[bot]@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 11:11:50 +01:00
David MannandClaude Opus 5.5 130b2b373f SVG: read the DOCTYPE's internal subset when checking the markup
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 11:08:24 +01:00
Federico Olivares b6d0f34500 Carry #9104 with #8888 and keep the raw strings of a rewritten Txt2
Merge #9104 (raw EngineData strings are written back as the same bytes) so a style-only type edit does not re-encode Photoshop's raw strings as UTF-16, with tests for the numbered and the legacy restyle paths. The over-budget pattern warning is written without a panic path and explained where the lock is held.
2026-10-25 07:06:54 -03:00
David MannandClaude Opus 5.5 bd25b8c805 SVG: an embedded data-URL SVG gets the nesting-depth check (#9397)
usvg parses an SVG embedded as a data: image (image/svg+xml, or text/plain
that isn't PNG/JPEG/GIF/WebP) with its default resolver, past the depth
check parse_with runs on the drawing, so 20 000 nested groups in it
overflowed the stack. The resolver now runs nesting_depth on the embedded
markup (after the bounded inflate for gzip) and drops the image past
MAX_XML_DEPTH.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 11:06:29 +01:00
highperfocused b81005280b fix(ui): correct the HUD hue strip mesh, marker, and right-erase precedence
CodeRabbit review of #6236:
- the hue strip's band triangles used b+3 (a vertex two rows ahead)
  instead of b+2, distorting every band and indexing past the last
  vertex on the final band; add a test checking every index against
  the vertex rows
- the hue marker's second endpoint was the strip's right center, so
  it drew a diagonal instead of a horizontal bar at the hue
- pointer_secondary opened the HUD on Shift+Alt+right even with the
  right-erase preference; the canvas gives that preference precedence
  (canvas_buttons' hud_gesture excludes erase strokes), so mirror it
  and add a control-channel test
2026-10-25 11:04:10 +01:00
David MannandClaude Opus 5.5 d21c06eee3 Integration: the Liquify wheel test turns Animated Zoom off for its one-frame notch (#5547)
The Liquify preview wheel now eases with Preferences › General › Animated Zoom
(on by default), like the canvas and Navigator, so a notch spreads over several
frames. The #7210 test checks the instant x1.1 step, so it turns Animated Zoom
off first, as the Navigator wheel test does.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 11:02:01 +01:00