Merge Swatches relinked a gradient stop linked to a merged spot swatch to
the kept swatch but kept the process plates it printed with the old ink,
so a kept spot swatch printed them beside its own tint and a kept global
process swatch held them for a later change back to spot. The merge now
clears those plates in the art, gradient swatches, graphic styles, the art
of brushes and the fill and stroke for new art, in the same undo step.
Stops linked to other swatches keep their plates.
File > Revert keeps the document's tab and uid, so a trace prepared
before the revert compared equal to one prepared after it when the
restored image, object and settings were the same. A background trace
that finished after the revert then applied to the restored content, and
a result handed to imageTrace.make or imageTrace.makeAndExpand stood in
for a fresh trace.
The trace inputs now record the document's revert count. A run prepared
before a revert is no longer current, the commands drop its result and
trace the restored image again as one undo step, and the status bar says
"Image Trace finished, but its document was reverted: trace it again".
Reverting another document, or a revert that fails, does not affect the
run.
Eraser Tool Options gives Angle, Roundness and Size each a variation menu
(Fixed, Random, Pressure, Tilt, Bearing, Rotation) and a Variation field.
path.eraseRegion takes angleMode, roundnessMode and sizeMode with their
variations, sweeps a varying tip as a Calligraphic nib with what the pen
read along the drag, and refuses an unknown mode or a variation out of
range. The Eraser records the pen's pressure, tilt and rotation when a
mode follows them, and its tip outline at the pointer takes the angle,
roundness and size the pen gives while it erases.
The canvas rendered its art at the display's pixels per point whatever
the canvas's size, so a canvas more device pixels across than egui's
max_texture_side asked for a texture larger than the display supports.
The tiled panes did the same, and Pixel Preview allowed regions of up to
16,384 document pixels. In
canvas::panes::tests::tiled_documents_fit_their_panes_at_every_ui_scaling,
the pass where a new UI zoom takes effect sizes the screen from the
default 10,000-point viewport, and the test failed whenever the render
worker finished that frame before the next pass.
- The canvas and the tiled panes lower the frame's pixels per point when
a side would pass max_texture_side, and draw the frame over the whole
canvas at that lower resolution.
- Pans, scrolls and scroll bar drags move the view by whole pixels of
that frame, so a pan still copies the pixels it moved.
- Pixel Preview shows the normal render when its region of document
pixels is larger than max_texture_side.
- Tests that drew a canvas wider than egui's default limit of 2,048
pixels now state the limit their display reports.
After Select All in a threaded story whose edited frame is empty, the
panel's commands style the text of every frame, but the checkbox rows,
Stylistic Sets and the East Asian items read only the empty frame. They
now read the runs of every frame, leaving out runs with no text while
any run has text, so a feature the story's text has shows checked and a
click turns it off.
Files saved for versions 10 to 13 place each symbol instance with a
TransformMatrix. On the page it takes the symbol's art to the canvas
(y down); in a symbol's art, it takes the art to that symbol's space. The
reader only knew the RTransformMatrix of newer files, so any such
instance made the editing data unreadable, and the file opened from its
PDF part alone, which holds only the art on the artboards.
The reader now converts a TransformMatrix with the canvas point of art
space's origin that the template box gives. A file whose symbol
instances were the only part it could not read now opens from its
layers, with the art outside its artboards.
The symbols fuzz test also damages the same data in the older form.
Docs: docs/target-app-parity.md, docs/file-format-parity.md.
With Output > Convert to Destination and an sRGB destination, a gray or
RGB image that carries an ICC profile (a Display P3 TIFF, a PNG or JPEG
with an embedded profile) is converted from that profile to sRGB with
the color settings' rendering intent and written without the profile.
Before, it kept its own numbers and profile.
- A 16-bit image that is not resampled or recompressed stays at 16 bits
a sample from PDF 1.5 on.
- A JPEG kept as JPEG whose samples the conversion leaves unchanged (an
sRGB profile) keeps its JPEG data; other JPEGs are encoded again.
- A black-and-white image whose converted levels are gray is compressed
as grayscale.
- A profile the color engine can't read is kept, with a warning.
- Without conversion, and with Preserve Numbers, images keep their
profile as before.
IccProfile::to_srgb_pixels converts a run of pixels in one pass.
Docs: docs/mcp.md, document.exportPdf params, target-app-parity.
An opaque knockout group in Normal mode with isolate off and nothing
blending inside was written as an isolated PDF knockout group (/I true
/K true), and a clipped one was drawn in an isolated form. The group's
elements now start from the art below it, as on the canvas: the /K
group carries no /I, and a clipped group gets no isolated form. An
isolated knockout group keeps /I true.
A click on a symbol instance hit only its art's geometric box, and a
marquee touched only that box, so art that a clipping path's live
geometry effects move past the box could be seen but not selected.
- Outside its box, in Preview, a click hits an instance where its placed
art hits (dynamic overrides and 9-slice placement included), and a
marquee touches the placed art as it touches a group of that art.
Outline view and layers with Preview off keep the box alone.
- The quick reach of an instance takes in the regions of the clipping
paths with live effects in its art, so the groups around it no longer
cull those clicks.
The editable text of an .ai save now lists the Hard and Soft Kinsoku sets
(resource key 4) and writes each paragraph's Kinsoku set (key 27, /nil for
None), Kinsoku Shori Type (25), Burasagari (24) and Bunri-Kinshi (23), in
every paragraph run, the normal paragraph style and the document's
defaults, in the form the .ai reader reads. The four settings reopen as
set, including Burasagari and Bunri-Kinshi under Kinsoku None, and are no
longer reported as left out.
Docs: docs/mcp.md (Bunri-Kinshi), docs/target-app-parity.md.
A gradient stop's process plates print beside the tint of the spot swatch
that the stop itself links to. A painted live effect could return a stop
with `plates` and no link, or a link to a process swatch or Registration,
and the result was drawn with plates that no plate ever prints.
- The painted-output decoder refuses `plates` on a stop with no `swatch`
(absent or null). The gradient's own `swatch` does not link its stops.
- The reference check refuses `plates` on a stop linked to a swatch that
is not a spot swatch. It runs again on a cached result, so a result is
refused once its swatch stops being a spot swatch. A missing swatch
keeps the error that names the swatch.
- Either fault refuses the whole result, and `lastError` names the
stop's `plates`. Plates on a stop linked to a spot swatch, all-zero
ones included, are kept. Filter and importer output keep clamping and
unlinking plates as before.
Docs: docs/plugins.md, docs/target-app-parity.md. The new message has a
row in every complete message catalog.
With linkPages and cropTo: crop, a PDF-compatible .ai opened on its crop
boxes, while opening it as art uses its trim boxes. It now uses the trim
boxes too. A page on its trim box shows its bleed box round the
artboard, and the pages' bleed (the widest on each side) becomes the
document's bleed, as the importer sets it.
linkPages read and checked the whole PDF again for every page it
linked. The pages are now read from the file opened once. With
cropTo: bounding, a blank page stopped the open with "has no art to
place"; it now gets its crop box, as opening the page as art does.
linkPages refused other files but let a .ai whose page is only a
placeholder through, and each such page then showed the file's editing
data. A page read from editing data is now refused with the existing
error. The file is also hashed once for all its linked pages instead
of once per page.
Extrude & Bevel, Inflate and Rotate decided whether a fill, stroke or
member casts part of the shadow with `opacity > 0.0` and skipped it with
`opacity <= 0.0`. A NaN opacity failed both tests, so the part was
neither cast nor skipped: the shadow cast from the solids drawn added its
faces with a NaN ink, giving the shadow fill a NaN opacity, and the
shadow built for itself recorded an empty part.
Both paths now use one rule, `shadow::Ink::casts`: art casts only when
its opacity is a finite number above zero. NaN and infinite opacity cast
nothing, as 0 % does, and the art beside it still casts. Finite opacity
casts as before.
- A form other than the default, set with text.setRangeStyle or
text.setFormat, takes away the characters' glyph choice, which would
otherwise keep drawing its own glyph. "default" keeps it.
- With a width form (fwid, hwid, twid, qwid), shaping leaves out palt and
turns kern and apkn off, and SVG output writes font-kerning: none and no
palt, as the OpenType feature list asks. proportionalMetrics and kerning
keep their values.
- The Glyphs panel menu counts a width form as available when a face has
the feature in GSUB or GPOS.
- At a caret, the menu's check marks and its no-op check read all the
text of the object, which is what the command formats. Choosing the
current form again is not skipped when a character has a glyph choice.
An action or journal entry recorded before Scale and Reflect followed the
turned bounding box has no basis value. Replayed, such a step now acts along
the page's axes, as it did when it was recorded; steps recorded now carry
basis "box" and replay along the box. The built-in Reflect Horizontal action
names basis "box", so it keeps reflecting across the box's own axis.
The symbol test that reflects a turned instance names basis "page", which
is the reflection it was written for.
Object › Distribute follows Object › Align and lists Horizontal
Distribute Left, Center and Right, Vertical Distribute Top, Center and
Bottom, Horizontal Distribute Space and Vertical Distribute Space. Each
item runs the button's `align.distribute*` command, so it uses Align To,
the key object and the Distribute Spacing value, and with the Artboard
tool chosen it spaces the selected artboards.
Unit::parse took one level of recursion per + - * / sign, so a text such
as "1+1+...+1" with 200,000 terms overflowed the stack and aborted the
process. Every number field, parse_number and every command parameter read
as a length went through it.
Text with more than 128 + - * / signs now reads as nothing; text with 128
or fewer reads as before. docs/mcp.md states the limit for length strings.
The test a_very_long_expression_reads_as_nothing aborted with a stack
overflow before the change.
At a Type tool caret the panel's commands style all of the text, but the
checkbox rows, Stylistic Sets and the H or V Style and Roman Italics
items showed only the caret's style. A click could then turn a feature
off for the whole text while most of it did not have the feature. At a
caret they now read every character of the text: a mixed row shows a
dash, a mixed item shows unchecked, and a click turns the feature on
throughout.
The PSB row of docs/file-format-parity.md says the merged image comes in
clipped by the file's clipping path, as the PSD row does, and
docs/target-app-parity.md's evidence list gets a line for it.
The measurement that runs before the style cascade and usvg now covers
the inputs that still got past it:
- usvg matches a selector without keeping what it found, so one whose
leftmost part matches nothing tests every set of ancestors in turn.
The measurement counts those tests for each copy and stops at
50,000,000; past that the SVG is refused with "the SVG's style sheets
take too long to match". An SVG image's matching stops at what the
SVG showing it left.
- A test counts one more for every 16 attributes a lookup scans and
every 16 bytes of the value it compares, so many classes against a
long class list count the work they do, in usvg's matching and in
the import's own cascade.
- The declarations the import's own cascade keeps, once per element,
count toward the 64 MB, including those of `:not()` rules, which usvg
drops. Each `:not()` alternative counts as a test.
- A linked SVG file is measured before its cascade runs. The linked SVG
files share one measurement: once they pass the limits together, the
files after that are shown without their links. The tests a file past
the limits made still count.
- SVG images share the element limit and the selector tests with the
SVG showing them, as they already shared the 64 MB.
The svg_filters reader measures its SVG before the cascade too.
A PSD or PSB file can name one of its saved paths (image resources 2000
to 2997) as its clipping path in resource 2999. psdread::clipping_path
reads that path in the image's pixel space, under the even-odd rule the
file format specification gives, with a subpath around the whole image
first when the initial fill record says the fill starts with all
pixels.
File > Place of such a file, and File > Open of its merged image, put
the image in a clip group of that path with no paint. file.place answers
the image's id as `image`, as for a DCS file. Replacing an image alone
in a clip group with such a file keeps the image's scale and rotation.
links.update and links.relink leave the clip group as it is. Files that
name no clipping path come in as before.
#7569 changed the Extrude & Bevel and Inflate check so that a filled
shape's solid is built once, about the bounds the effect draws it
about, and stated that a filled shape whose solid builds about those
bounds but not about its profile's own bounds now passes. No test
covered that case.
The new test builds a group of a filled shape of 900 small curves and
a line with a 2000 pt stroke. About the shapes' own bounds, the filled
shape's solid is too complex with a repeated rounded bevel; about the
bounds that take in the stroke, its curves are sampled more coarsely
and it fits. The test asserts the first result, then that the check
passes and the solid draws. It fails when the check builds every
shape's profile about the profiles' bounds again.
The Rough Pastels dialog has a Load Texture… button, as Glass and
Texturizer have. The image's gray levels become the paper surface,
stored with the effect as `textureImage` and listed as Loaded Texture
in the Texture menu; Scaling, Relief, Light Direction and Invert act on
it as on the built-in surfaces. The dialog's Texture and Light
Direction controls are now menus instead of text fields.
effect.apply and effect.setParams read `textureImage` from a file or
take the stored string and refuse anything else, and a plug-in filter
cannot have the host read a file for it.
In the Effect Gallery's Settings, a file dialog that answers later
(the macOS sheet, the Linux dialog) now gives the texture to the values
that asked: the filter being browsed, or the stack row that asked,
found by a number each row now carries. An answer for a removed row, a
filter no longer browsed or a gallery that closed is dropped.
Params text, docs/mcp.md, docs/control-protocol.md, docs/gaps.md and
the evidence line are updated.
Type on an open path can now start before the path's start: its start
is stored below 0, the span runs from there, and the characters whose
middle falls before the path's start are not drawn, as SVG 2 lays out
text on a path (https://www.w3.org/TR/SVG2/text.html#TextLayoutPath).
The text keeps every character, and the start bracket stays at the
path's start. Past the path's end, text that does not fit is overflow,
as for other type on a path. Type that starts on the path draws every
glyph as before.
- SVG import keeps a <textPath> start before an open path's start
(text-anchor end or middle, or a negative startOffset) instead of
starting the text at the path's start, which drew every character.
- SVG export writes it as a negative startOffset.
- PDF export writes only the characters drawn on the path.
- The legacy .ai and .ai editing data writers write the type from the
path's start with the type attributes warning.
- type.pathOptions takes a start below 0 for type on an open path; on a
closed path it is the path's start, as before. A flip starts the type
at its start bracket. Dragging the center bracket slides it from where
it starts, and no further before the path.
- A click at the path's start places the caret by the first drawn
character, and the caret in the hidden characters stands at the
path's start. The highlight of substituted fonts and glyphs leaves
them out.
- Release Type on a Path keeps the first drawn glyph where it was
drawn, with the hidden characters before it.
- The Touch Type tool shows no box for a hidden character, and
text.touchType returns an error for one. A character with a drawn
glyph is changed by that glyph.
Docs: docs/mcp.md, docs/target-app-parity.md.
With Noto Sans Arabic the fatha's ink reaches past the first letter's
left edge, so that letter's Tsume only narrows its advance and moves no
glyph. The test compared glyph positions alone and found no change.
It now compares each glyph's position and advance.
Opening an .ai graph now sets placed_art from the art as read (main's d3789f2b0c), so the comparison in the graph round-trip tests leaves it out, as it already leaves out placed.
The menu item checked a listed script and then ran script.run by its path,
which opened the file again; a link put in place of the file or a folder on
its path in between could make it run a file the check never saw, and the
include folder was looked up from the path again as well.
script.run takes installed: true. With a path it reads the file once through
script::folder::read, runs that text, and reads #include and $.evalFile in the
Scripts folder's real path followed by the names that read found. The menu
item runs script.run {path, installed: true}.
Danish joined the complete-message languages on main, so the script.run answers errors and reasons now need da rows. The merge of main also left the indexed wrapper twice in uk and vi; the duplicate rows from this branch are removed.