File ▸ Close, its shortcut (Ctrl/Cmd+W) and the tab's × closed a document
with unsaved changes at once. Closing also discards the document's
recovery data, so the edits were gone for good.
The three user paths now go through menus::close_document. A document
with unsaved changes opens a dialog with Save, Don't Save and Cancel:
- Save runs the same Save the menu does, then closes. If the document
is still unsaved afterwards (the file picker was cancelled), it stays
open.
- Don't Save closes it.
- Cancel (or Escape) does nothing.
A document with nothing to save closes as before. The engine command
file.close is unchanged and never asks, so scripts, MCP and the control
channel's engine.execute keep working unattended. ui.menu.invoke without
params acts like the menu item, as it does for other dialogs.
Quitting the app is not covered here. Recovery data survives a quit, so
unsaved documents come back on the next start.
Co-authored-by: Claude Opus 5.5 (1M context) <[email protected]>
Co-authored-by: Brandon Thomas <[email protected]>
A placed PNG, GIF or WebP that is cut off (an interrupted download) or
has a bad checksum made the whole PDF export fail with
"PDF writer error: Image(Image(ImageRepr {..}), None, "unexpected end of
file")", which names no file. A damaged TIFF or BMP in the same place is
skipped with the warning "image `name` could not be decoded and was
skipped", and the rest of the document is exported.
The exporter decides whether an image can be embedded by asking krilla
to open it. krilla opens PNG, GIF and WebP lazily: it reads the header,
accepts the file, and decodes the pixels only when the PDF is written,
where an error ends the export. TIFF and BMP are decoded by us first, so
they took the skip path.
lossless() now decodes the image first for every format. A file that
does not decode returns None and takes the existing skip-with-warning
path. This costs one extra decode per PNG, GIF or WebP asset: the sample
magazine (two PNGs, 820 KB) exports in 0.19 s instead of 0.17 s in a
debug build.
Co-authored-by: Claude Opus 5.5 (1M context) <[email protected]>
Co-authored-by: Brandon Thomas <[email protected]>
In every workspace that shows the Control panel (Advanced, Typography,
Printing and Proofing, Book), the second row of fields was cut off at the
panel's lower edge: Y and H in object mode, leading, tracking and the
indents in text mode.
The panel lays its groups out in a row that centres each child
vertically. The two-row grids were placed as if they were one row tall
and then grew downward, so they started 13.5 pt below the panel's top
instead of at it and ended 5.5 pt past the 59 pt panel.
The row is now top-aligned, which is where the font menus, the
separators and the reference-point proxy already sat.
Co-authored-by: Claude Opus 5.5 (1M context) <[email protected]>
Co-authored-by: Brandon Thomas <[email protected]>
Placing an RTF or Tagged Text file dropped every character outside the
Basic Multilingual Plane: emoji, mathematical letters and the rarer CJK
ideographs such as 𠮷 (U+20BB7, used in Japanese names). The text
around them came in, so the loss was silent. It also broke our own round
trip: Export Text writes these characters correctly and Place lost them.
Both formats carry such a character as a UTF-16 surrogate pair, two
units in a row: \u-10179?\u-8704? in RTF, <0xD83D><0xDE00> in Tagged
Text. Both readers passed each unit to char::from_u32 on its own. Half a
pair is not a character, so each half was dropped.
A small shared Utf16Units joins the units: a first half waits for the
second, and the pair becomes one character. Half a pair with no partner
is still left out, as before.
Co-authored-by: Claude Opus 5.5 (1M context) <[email protected]>
DrawCraft was renamed VectorCraft, and its repository is now
storytold/vectorcraft. Three places still used the old name:
- AGENTS.md and CLAUDE.md told contributors and agents that the sibling
checkout lives at ../drawcraft. A clone made from the README link lands
in ../vectorcraft, so that path does not exist.
- The About window's sibling list used the app-page slug "drawcraft".
getartcraft.com/apps/drawcraft still redirects to /apps/vectorcraft,
so the link worked, but the README (which the list says it matches)
uses /apps/vectorcraft.
Co-authored-by: Claude Opus 5.5 (1M context) <[email protected]>
In the browser, pressing AltGr while editing text on the canvas inserted
the word "AltGraph" into the story.
eframe's web backend (0.36.2, web/input.rs text_from_keyboard_event)
turns a keydown into a text event unless the key is on a fixed list of
names. Browsers report many more named keys than that list holds:
AltGraph, Dead, Unidentified, the media keys. Each one arrived as text.
Panel text fields are not affected, because a focused egui text field
takes its text from the browser's input events instead of keydown.
The web shell now drops a text event that is a key name (ASCII letters
and digits, starting with a capital, longer than one character) unless
an egui text field has the keyboard. One key press never types such a
word; a text field's input events can, so they are left alone.
Co-authored-by: Claude Opus 5.5 (1M context) <[email protected]>
Three commands took a count from their parameters and used it unchecked.
Parameters come from MCP, the control channel and scripts, so the count
is hostile input (never-crash: cap input-sized allocations, checked
arithmetic).
- layout.createGuides made one or two guides per row and column with no
upper limit: 30 million rows took 1.2 GB, and 2^32 - 1 would exhaust
memory, which aborts the process past the command guard. A count over
2^32 wrapped to a small one. Both counts are now capped at 100.
- edit.stepAndRepeat multiplied rows by columns as usize. In debug builds
the overflow panicked; in release builds 2^32 x 2^32 wraps to 0, passes
the 1000-copy limit and starts a grid of 2^64 copies. The product is
now checked, and a grid that large falls back to `count` copies like
any other grid over 1000.
- file.new stored any column count in the page (4 billion, or a wrapped
value past 2^32). Page::column_rects builds one rect per column each
time the columns are laid out. The count is now clamped to 1..=216,
the range layout.marginsAndColumns already uses.
Each case has a regression test that failed before the fix.
Co-authored-by: Claude Opus 5.5 (1M context) <[email protected]>
Rust 1.98 adds the clippy lint chunks_exact_to_as_chunks. It flags two
loops in the PSD decoder, so `cargo xtask ci` stops at the clippy step on
unmodified main for anyone on current stable.
Both loops now use as_chunks / as_chunks_mut (stable since 1.88, below the
workspace rust-version 1.90). The chunks are arrays, so the try_from that
could never fail goes away too. Decoded pixels are unchanged; the existing
PSD tests cover both paths.
Co-authored-by: Claude Opus 5.5 (1M context) <[email protected]>
