Reverts the fallback that merged the layers of art selected on the canvas when no
layer row is highlighted: Illustrator's Merge Selected merges the highlighted rows
into the last one highlighted.
sRGB to Generic CMYK applies black-point compensation (black separates to the
darkest rich black) but CMYK to display did not undo it, so hex 000000 became
C67 M67 Y67 K100 and then showed as #1b1818. The display XYZ now maps the press
black to black when the colour settings use black-point compensation, in the
sRGB conversion, the ink lookup table the canvas uses, Lab of a CMYK colour and
the gamut warning.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The overflow marker was drawn by the Type tools only, so with the Selection tool a text box smaller than its text showed no sign that text was hidden. Session::overlays now adds the marker of every selected area type and type on a path that overflows, unless a Type tool is active (it draws its own). The verdict is cached per object and font generation.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
With no highlighted rows and art selected in two or more layers (Select All), layer.merge merges those layers; the panel menu enables it then.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Multiply squared a colour overprinting itself (stroke over same-colour fill showed a darker band). Darken keeps the stronger ink per plate/channel.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The language the UI is drawn in was one process-wide value in unit
tests too, and the tests run on parallel threads.
current_language_is_a_frame_setting sets Traditional Chinese for a
moment, and a test that drew text in that moment read it in Chinese.
chrome::tests::the_zoom_hint_follows_animated_zoom failed that way in
29 of 500 runs together with the i18n tests.
The drawing language is now one per process in the app and one per
thread in unit tests, as ui_mirror! keeps the shortcut and workspace
mirrors. The app keeps the same static. A new test sets the language
on another thread and checks that its own thread still draws in
English.
The tmp() helpers in apps/vectorcraft-cli/tests/cli.rs and fileio.rs
name their folder after the process id (vectorcraft-cli-test-<pid>,
vectorcraft-cli-fileio-<pid> in the temp folder) and never empty or
remove it. When a later run gets the same process id, it finds the
earlier run's files there. mcp_automation_roots then fails on Unix with
EEXIST (Os { code: 17, kind: AlreadyExists, message: "File exists" }):
its first symlink() call finds the link in/escape.svg that the earlier
run made. A stale output file can also let a fileio test read an
earlier run's export.
The first tmp() call in each test process now removes the folder and
creates it again empty. The roots test's body moves into
automation_roots(), and mcp_automation_roots_in_a_folder_an_earlier_run_left
runs it twice in one folder, emptying the folder in between as tmp()
does. Without the removal the second run fails with the same error.
On main, cargo xtask wasm printed an unused-import warning for json in
crates/engine/src/cmd/fileio/action_batch.rs, and this check reported
it as an error:
RUSTFLAGS=-Dwarnings cargo check -p vectorcraft-engine --target wasm32-unknown-unknown
Every use of json! in action_batch.rs is in code compiled only for
native targets. The wasm32 version of action_batch uses Value only.
The json import now has the same #[cfg(not(target_arch = "wasm32"))]
attribute as the code that uses it. Native builds import the same
names as before.
Found while testing crates/doc/src/xml_id.rs:31 (from_object_name) / :10 (to_object_name)
Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Found while testing crates/engine/src/cmd/graph.rs:193 (fmt_value)
Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
document.rasterEffectsSettings accepts addAround from 0 to 1,000 pt,
but a file's raster_effects.addAround opened as it was saved. With
addAround 1e300, the area of a raster effect image overflowed f64,
the image scale became 0 and the image got the transform
[inf, NaN, NaN, inf, x, y]. PDF export left that image out, so the
drop shadow was missing.
A file's addAround now opens clamped to 0 to 1,000 pt. An effect
image whose area or scale is not finite is not made. Export then
keeps the object with its effects, and Expand Appearance expands it
without an image, as when its PNG can't be encoded.
Docs: docs/mcp.md (addAround).
Before the Eyedropper samples an image, or Edit Colors, Recolor Artwork
or a color adjustment effect recolors an embedded one, `rgba_within` reads
the image's size from its header and checks its size as RGBA against the
allowance. The check multiplied width × height × 4 in 64 bits. A TIFF's
sides go up to 2^32 - 1 pixels, so a header of 2^31 × 2^31 pixels
overflowed that product: debug and test builds panicked with "attempt to
multiply with overflow", and release builds wrapped it to 0 before the
decoder's own 512 MB limit refused the image.
The check now compares width × height with the allowance divided by 4.
Every size whose RGBA bytes fit in 64 bits gets the same answer as
before, and a larger one is refused.
A raster export compared the process-wide count of images that could
not be decoded before and after it drew, and warned that "an image
couldn't be decoded and is missing from the picture" when the count
rose. Any other thread that failed to decode an image in that time
raised the count too: a second export, the UI thread drawing the
canvas, or the canvas, thumbnail and placed-document workers drawing a
document with such an image. The export then warned about an image it
does not contain. In the engine's tests,
options_not_applied_yet_and_document_features_warn failed in 5 of 60
runs of the tests_raster and tests_pdf* tests.
The count is now kept per thread. An export reads it on the thread it
draws on. That thread decodes every image the export draws that is not
already in the shared image store, and an image that failed to decode
is never stored.
The renderer decoded each image within its 2 GB allowance (512 MB in
the web build) and then converted it to RGBA. The decoder holds the
decoded image, before conversion, to that allowance, so an image stored
with fewer than four bytes a pixel (gray, 1-bit, gray with alpha, or
RGB without alpha) could take up to four times as much (4/3 times for
RGB) once converted, and as much again as the pixmap drawn from it.
The renderer now reads the image's size from its header first. When
width × height × 4 is over the allowance it does not decode the image
and handles it as one it can't decode: the renderer draws nothing for
it, and a raster export warns that an image is missing. Images within
the allowance are decoded and drawn as before.
Crop Image's trim (object.cropImage {trim: true}) decoded the image
within MAX_RASTER_ALLOC (2 GB) and then converted it to RGBA. The
decoder holds the decoded image, before conversion, to that allowance,
so a gray or 1-bit image within it could take up to four times as much
once converted.
Trim now reads the image's size from its header first and, when
width × height × 4 is over the allowance, fails before anything is
decoded, with the check and message Load Texture uses: "an image of
W × H pixels is too large to draw". Images within the allowance are
trimmed as before.
BFlat