An option given to document.exportSelection skips the options dialog, but
without `path` the save panel still picks the file. docs/mcp.md said any
option writes without a dialog, and docs/control-protocol.md said params
export without the dialog. Both now say that only `path` writes without a
dialog, and a test pins the behavior.
- appearance.moveItem with from "contents" and aligned: true is refused when
the objects' stacks differ, as it is for rows.
- paint.invert and paint.complement with an item remember the paint of an
object they recolored, not that of design art they left out.
- effect.remove and effect.duplicate with indices and aligned: true are an
error, and change nothing, when the design art given first has no effect
at one of the indices.
oversized_tiff now also refuses a TIFF whose JPEG-compressed strip or
tile (Compression 7) declares a frame wider or taller than the strip or
tile, one with more than MAX_TIFF_CHUNKS strips or tiles, and one whose
JPEG headers take more than 16,777,216 steps to read. It reads BigTIFF
directories, follows the layout and the repeated tags the decoder
reads, and refuses offsets of types it does not read.
Keep upstream codec and Global Edit work identical to main so only the 38 Tech Circuit feature files remain in the PR diff. Resolve the G5 documentation conflict by preserving completed Global Edit behavior and the Tech Circuit estimate reduction. Fresh deep CodeRabbit review of the exact feature diff passed with zero findings; no full local rebuild repeated.
When Merge added a path that cut a painted edge, the piece holding the
old edge's middle took the old edge's stroke rows unchanged, so its
dashes started again and it kept both arrowheads. It now takes them
fitted to its part of the old edge, as Gap Options does.
The prepass writes a key in place of each <feImage> data URL before usvg
reads the SVG, and the filters kept as Effect > SVG Filters were read from
that edited text, so their XML held the key. The effect's check refused it
and the filter was dropped with a warning. The filter's XML now has each
keyed href put back as it was written.
The check before a PNG or JPEG is embedded as it is read the profile's
color space only from a profile of at least 128 bytes, so a gray or RGB
profile cut short after its color space reached the PDF writer. The
color space is now read from bytes 16..20 of any profile that holds
them, and valid_icc rejects the short profile.
When the last of the bounded breaks such a line reads came right before
U+2028, the line ended at that break and the next line held only the
forced break, drawing an empty line. The line now reads on to the
forced break and ends there when it fits as drawn.
In PDF/X-1a, a tagged gray JPEG that the Compression settings keep as
JPEG was decoded and written with Flate. It is now written with its
JPEG data unchanged, in DeviceGray without its profile.
A PNG or JPEG embedded as it is now has its gray or RGB profile checked
with valid_icc, like the images the recoder writes: a malformed profile
is left out, the JPEG keeping its JPEG data and the PNG decoded.
After the merge of main, the tracked-ligatures note is the only reader of
the first run's features, and the branch replaces that check with one that
reads every frame of the story, so clippy reported the binding as unused.
main added the same 13 custom-toolbar `@msg` rows to ja.tsv and hu.tsv,
so the branch's copies made `bundled_catalogs_are_consistent` fail with a
duplicate key. Keep main's rows and remove the branch's.
tab_advance read the text after a tab up to the next tab for every
stop, though only right, center and decimal stops align it. The first
tab of a long paragraph read the paragraph to its end, even on a line
that stops reading a bounded number of breaks past its overflow. It
now reads that text only when it reaches a stop that aligns it.
A closed path that crosses itself is cut into closed loops. The loop
holding the path's first anchor counted as the whole path, so it kept
the path's width profile and brush. A closed edge now counts as the
whole path only when it is as long as the path.
A whole pick of every fill and stroke attribute, or an append pick, onto
a graph's axis line, tick mark or radar ring stored the line's rows in
full, object opacity and blend mode included, even when the pick did not
apply the object's transparency. On a line that still showed 100% until
the graph was drawn again, that replaced the opacity its kind kept (40%
set on another line became 100%). An append pick onto a graph label, and
graphicStyle.apply with add on a label or line, did the same.
Without Transparency applied, these picks, and a style added on top, now
store the fill and stroke rows and leave the kept object opacity and
blend mode as they were.
A gray or RGB image is tagged with the ICC profile it carries only when
the profile passes the checks a CMYK image's profile already passes: at
most 16 MiB, the size its header declares, the acsp signature, an input,
display, output or color space profile class, and a tag table that lies
within it. Otherwise the image is written in DeviceGray or DeviceRGB, kept
as it is, resampled or compressed with JPEG.
When the third click of a triple-click landed on a face or edge that
the double-click before it had painted (a neighbor the flood or the edge
walk reached), the triple-click sent that part's new paint as match,
since only the clicked part's paint was recorded. The bucket now records
the fill or drawn stroke paint of every other part a double-click
reaches, and a triple-click on one of them matches the paint it had
before the double-click. docs/mcp.md says so.