Conflicts: canvas.rs and dispatch.rs keep main's Gradient Overlay drag and
sampler release_taken_drag next to the preview locator; cursors.rs keeps
main's cached Paint Bucket cursor beside preview_locator_cursor; ui-parity.md
keeps both status lines and revision rows.
Co-Authored-By: ljkx <ljkx@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
main added preferences since the PR's last merge, so the toolColors field now
makes it 10 of 195 settings that do nothing (was 10 of 194 on main); the
ui-parity preference row follows the scorecard.
Co-Authored-By: GyoJoong Kum <david32123kr@gmail.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Create Droplet (CrtD) loads as file.automate.createDroplet for the form
the droplet plays the same way (a Folder destination with Override
Action "Save As"), plays, and saves back as recorded; subfolders, Save
and Close, custom naming, an error log or suppressed dialogs keep the
Photoshop step. The Gradient tool's Make + Set of a gradient fill layer
already map since #10771; a test now covers that recorded form. Print
stays a kept step: its two recorded steps would print twice.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Main moved discard_ui.rs to dialogs/discard_ui.rs and its tests to
dialogs/discard_ui/tests.rs; the change and its tests now live there.
Co-Authored-By: devstivi <devstivi@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Main's slider rows now name their slider after the caption and link the caption to
the value field, so get_by_label("Scatter") found the field instead of the caption.
The tests use caption_rect, and scatter_row names its slider like slider_row does,
with a test that fails without it.
Co-Authored-By: augrclk <augrclk@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Main moved brush_sections.rs to panels/, its tests to brush_sections/tests.rs and
split widgets.rs into modules. The PR's tests now live in brush_sections/tests.rs and
scrub_label_within is crate-visible from widgets/scrub.rs; both doc status lines and
history rows keep both sides.
The Type tool's default font size is 50 pt, not 48 pt. Measured in the
reference application: after Reset All Tools the size field reads 50 pt,
the same in an 800x600 document at 72 ppi, a 3000x2000 document at 300
ppi and a 100x100 document at 72 ppi, and the scripted tool options
report size=50 in points.
`type_size` had one place where the number lived, so a fresh install,
Reset Tool and Reset All Tools (`tool_reset::reset_options` /
`reset_all`) and the fallback for a damaged remembered value
(`tool_memory::sanitize`) all follow the same constant. Tool options
saved by an older version are still restored field by field, so a
saved 48 pt stays 48 pt.
Tests go in the new `state/type_default_size_tests.rs`. `state.rs` only
gains the `mod` line the test file needs.
Main moved the files this PR changed (engine/src/cmd/edit_menu_cmds/fade.rs,
ui-egui/src/canvas/filter_preview.rs, dialogs/filter_dialog.rs) and Fade now also
fades alpha channels and the Quick Mask. The preview is ported onto that:
- engine: fade_preview captures the blend (FadePreview) from the session; render()
blends it on a copy that shares the document's tiles, so only the faded area is
blended and rewritten (review: no whole-surface copy).
- ui-egui: the blend and the flatten run on the filter preview worker in the
desktop app, like every filter preview, and inline on the web.
- tests: the preview matches OK for alpha channels and the Quick Mask too; the
dialog test previews, cancels (document, history and Fade unchanged), then OKs
on both the worker and the inline path.
Co-Authored-By: Korg1ch <Korg1ch@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
main moved brush_sections.rs to panels/ with its tests in
panels/brush_sections/tests.rs, and gave the Angle and Roundness fields
accessible names. Keep the PR's left column and right-hand widget with
main's accessible names, move the PR's three layout tests and
section_harness_width into the new tests file, and find the captions with
main's caption_rect (get_by_label now also matches the named fields).
Docs: both sides' status entries kept.
Co-Authored-By: augrclk <217665348+augrclk@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Measured in the reference application: carrying an open Curvature Pen
path on from its endpoint gives exactly the handles of a path drawn
from that endpoint on. The endpoint keeps its incoming handle and its
outgoing one is the new run's first handle, so two arcs meet in a cusp.
The PR bent the outgoing handle in line with the incoming one instead,
which sent the new segment off the other way. The first knot's incoming
handle on closing still stays in line with its outgoing one, as there.
Co-Authored-By: Lcoding15 <125505497+Lcoding15@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The fields rounded a typed length to the unit's readout precision on lost
focus ("12.5 px" became "13 px"), and a bare number always took the ruler
unit. The reference application keeps three decimals in every unit (20 px
at 72 ppi opens as "0.706 cm", "13.5 px" stays) and gives a bare number the
unit the field showed (after "2 mm", "5" becomes "5 mm"). Also merges main
(dialog moved to dialogs/, accessible names kept) and folds the two length
field widgets into one.
Co-Authored-By: safvanedge <241307593+safvanedge@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Main has meanwhile carried on open shape-layer and vector-mask paths
(754c61fd7f) and moved vector_ui.rs to tools/vector_ui/. The PR's own
continuation code in commit_path and resume_at is superseded by main's
and dropped; what is left of the PR is the continue-path pointer:
continues_path_at moves to tools/vector_ui/pen.rs next to
closes_path_at, and its canvas arm goes into main's split cursor match.
Co-Authored-By: kavishkashehan726-sudo <kavishkashehan726-sudo@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The reference's source-over computed s * sa / sa over an empty backdrop,
one ulp off the stored ink. Behind a Paint.NET Color Burn of a Posterize
that lands exactly on 73.5/255, the ulp flipped the 8-bit rounding and
random_documents_match_the_reference failed on main. Keep the source ink
when nothing shows through or both inks agree, as cmyk::normal does.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>