A file can hold any number of saved selections; the duplicate-name check and the name lookup were quadratic.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
(cherry picked from commit 9bcb695ad3904565ed84b7562aa99dd818604cec)
Choosing a perspective tool shows the grid, but the canvas also re-selected the tool after a
Cmd-drag with the borrowed Selection tool, so a grid hidden with the tool chosen came back after
any Cmd-drag. `Session::switch_tool` (and the app's `restore_tool`) switch without choosing; the
temporary-tool return uses it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The grid was drawn whenever its `visible` flag was set or a perspective
tool was active, and choosing the Perspective Grid tool (Shift+P) didn't set
the flag. So with the tool chosen, View > Perspective Grid > Hide Grid
(perspective.grid.show) turned the hidden flag on instead of off and the
grid stayed on screen however it was toggled.
Choosing a perspective tool now shows the grid (sets `visible`, view state,
no undo step), and the grid, its widget and the Perspective Grid tool's
handles are drawn only while it is visible. Hide Grid hides it with the
tool still chosen; choosing the tool again shows it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Parity: Object › Path › Remove Anchor Points now runs the new command (`path.removeAnchors`,
the menu's feature name) instead of `path.deleteAnchors`, which deleted the anchors with their
segments and opened closed paths; that stays the Delete key's command ("Delete Anchor Points",
off the menu). No second "Smart" item.
- Geometry: the refit moves to `vectorcraft_pathops::remove_anchor` on the existing least-squares
fitter (fixed end tangents, a golden-section search for where the curve passes the removed
point) instead of a ~1,000-candidate grid search in geom (about 4 ms per anchor, now well under
0.1 ms). Two straight sides stay one straight side (a rectangle loses a corner and becomes a
triangle), as Illustrator does; non-finite input leaves the handles alone.
- The Delete Anchor Point tool (`path.removeAnchor`) uses the same function, replacing its
heuristic handle stretch; the explicit id/anchor and `fit` params of the new command are dropped
as duplicates of it. The command is enabled by direct-selected anchors (`has_anchors`), not a
UI override.
- The Control bar shows an icon button (the Delete Anchor Point icon, tooltip "Remove Anchor
Points") instead of a 204 pt text button; the labels reuse the translated "Remove Anchor
Points"; zh-hant gains "Delete Anchor Points". The Pen test that pinned Alt-click as plain Pen
(Illustrator's Alt over an anchor is Convert Anchor Point) is dropped.
- Tests for the refit, closed paths, straight sides and bad input; docs/mcp.md and ROADMAP.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
path.smartRemoveAnchor stays. Object › Path, the Control bar, the contextual task bar, Properties, and the canvas context menu run it on direct-selected anchors. The Pen no longer uses Alt/Option for this, so that key stays free for Convert Anchor Point.
- A dragged handle no longer snaps to its own path's bounds and centre, which move with the
handle; the path's anchors are point and alignment targets instead.
- Shift steps 45° from the Constrain Angle preference, like the cut and gradient tools.
- snap_draw and snap_handle share one pixel/grid/smart-guides switch.
- A test that a smooth anchor's opposite handle turns with a Shift-dragged handle and that Alt
still breaks it, and the tools' module docs (both from #338).
Co-authored-by: theironchef <6278036+theironchef@users.noreply.github.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Dragging a direction handle with Direct Selection or the Anchor Point tool
sent the pointer position straight to path.setHandle: Shift did nothing and
the handle never snapped, while the Pen already constrains the handles it
pulls out.
guides::snap_handle places a dragged handle: with Shift at a multiple of
45° from its anchor (the Pen's constraint, keeping the dragged length),
otherwise snapped as a drawn point is (smart guides, the grid, pixels).
Both tools use it and show the smart guides while the handle moves.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
artboard.move with copy and artboard.duplicate now make the copy with one helper: a
new id that can't overflow and a name of its own (`<name> copy`, then `<name> copy 2`…,
so two copies of one artboard no longer share a name). `moved` lists the copies a copy
moved rather than the originals it left. Test it with and without moveArt, and note
Alt-drag in the ROADMAP.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Alt-dragging an artboard moved it like a plain drag. artboard.move now
takes copy: it leaves the artboard (and the art it carries) where it
is and moves a copy, named "<name> copy", with copies of that art. The
Artboard tool previews it while Alt is held, as "Duplicate Artboard"
(Alt pressed or released mid-drag switches between copy and move),
and makes the copy the active artboard, as in the reference app.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The engine kept its own copy of Document::next_layer_name after #353; the
document's now counts sublayers too, so the engine uses it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A new sublayer was named "Layer {node_count}", counting every object, so
an empty document's first sublayer was "Layer 1" like its parent and a
document with five objects got "Layer 6". Top-level layers counted only
top-level names and could repeat a sublayer's. Document::next_layer_name
now counts every layer, sublayers included, and skips any layer's name;
layer.newSublayer uses it too.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Between paths, the Pen now adds an anchor where a click lands on a
segment of a selected path and deletes the anchor a click lands on,
with the add and delete cursors while hovering, as in the reference
app. The ends of a selected open path still continue it. Shift held,
or General > Disable Auto Add/Delete (a stored preference nothing read
until now, passed to tools as ToolContext::auto_add_delete), starts a
new path instead.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Japanese full-width punctuation is half a mark and half a space (an
opening bracket's space before it; a closing bracket's, comma's or full
stop's after it). Set solid, 」「 or 。」 show a full em of air between
the marks, and a line ending in 。 stops half an em short of the edge.
- `ParaStyle.mojikumi` (Paragraph panel › Mojikumi): None, or Line-end
Punctuation Half Width, which (JLREQ 3.1.4) lets consecutive
punctuation share one half-em space — a closing mark followed by
punctuation loses the space after it, an opening bracket after another
the space before it — and sets a closing mark ending a line half
width. Only full-width marks (advance an em) are touched, never inside
a tate-chu-yoko block; vertical type the same way down the column.
- New type (`text.create`, Area/Path Type) takes Line-end Punctuation
Half Width, the usual default of Japanese layout; documents from
before it and imported text keep None (written only when set, so
older files and older apps are unaffected). `text.setFormat
{mojikumi: "none"|"lineEndHalf"}`.
- Builds on #344: on justified lines the room it frees is spread
between the characters.
- Catalog rows in ja and zh-hant.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Art in a sublayer is now selected as objects (clicks, marquee, Select All, the
Select menu), sublayers select in their own colour, and hidden, locked or
template sublayers block clicks. Layer Options gain Preview (per-layer outline)
and Dim Images. New and fixed layer.* commands for the Layers panel: setCurrent
and highlight (rows at any depth), selectAll, move (reparent with cycle and
nesting checks, Alt copies), delete/duplicate rows, merge, flatten, collect,
release to layers, reverse, template, hide/outline/lock others and their
show/preview/unlock all, locate.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
On hybrid-graphics laptops (an integrated GPU driving the display plus a
discrete one), egui-wgpu's default power preference is HighPerformance, so
VectorCraft rendered on the discrete GPU and its frames were copied to the
display through the integrated one. On an Asus ZenBook with Intel Iris + Arc
the window flickered whenever the mouse moved over it (#306, #305).
The canvas is rasterized on the CPU (vello_cpu) and the GPU only composites
the canvas texture and draws the UI, so the integrated GPU is plenty. The
desktop app now asks wgpu for the power-saving adapter by default.
Single-GPU machines are unaffected.
- Preferences > Performance > Graphics Processor (`gpuPreference`:
powerSaving, the default, or highPerformance), with a note that it
applies after a restart. The app reads it from the saved preferences
before the window opens. WGPU_POWER_PREF still overrides it.
- Help > About and `ui.inspect` (`graphicsAdapter`) name the adapter in
use, and it is logged at startup, for future GPU bug reports.
- Docs (README, docs/development.md) and ROADMAP updated.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
#249 made the reduced repros of #234 open, but a real file saved by the
reference app (version 30) still fell back to its screen-resolution preview
with "typecheck in `a dictionary`". Running its prolog to the end needed:
- The standard dictionaries (systemdict, userdict, globaldict, statusdict,
errordict, $error) as values in systemdict, so `/globaldict where
{ /globaldict get begin } if` gets a dictionary, not an operator.
SharedFontDirectory names the font directory too. errordict has a
handleerror for prologs that wrap it.
- `dictstack`, `internaldict`, `currentsystemparams` and
`currentuserparams` (with the cache sizes programs `get`), and `cshow`.
- Executable files: `cvx` and `cvlit` set a file's attribute, `xcheck`
reads it, and executing one runs its data as a program, as the prolog
does with `(...) 0 () /SubFileDecode filter cvx exec`.
- SubFileDecode with a count of 0 and an empty end string passes its
whole source through instead of nothing.
- `flushfile` on a filter over the program's data skips that data (the
file's XMP packet). A filter chained from a pending one leaves it
nothing to skip, so the writer's own `VCsrc flushfile` is unchanged.
The interpreter empties its standard dictionaries when it finishes, since
systemdict now holds itself. Both real files tested now import as 15
vector paths in CMYK with no warnings, and round-trip through SVG and PDF
pixel for pixel. The EPS fuzzer exercises the new operators.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A live rectangle kept every transform in its `xf`, so a non-uniform scale
(a bounding-box drag, Object > Transform > Scale) stretched the rounded
corners built in untransformed space into elliptical arcs. Scale Corners
off divided the radii by the mean scale, which only compensated uniform
scales.
LiveShape::transform now folds a scale along the rectangle's own sides
(any rotation or reflection, no shear) into w/h and keeps only rotation,
reflection and position in `xf`, so the radii are document lengths and the
path is regenerated with circular corners. Radii scale by the mean scale
(sqrt|det|), which keep_corners undoes for Scale Corners off. Shears and
plain moves keep composing into `xf` as before, so files saved with a
scale in `xf` keep their exact geometry until next scaled or rotated.
A corner-widget drag starts from the radius as drawn (clamped to fit).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Save Selection… asks for a name and keeps the selection in the document (saved with
the file, undoable, at most 25) instead of in session memory keyed by the title.
Each saved selection is listed at the bottom of the Select menu and a click selects
its objects again. Edit Selection… opens a dialog with the list to rename or delete
them, applied together on OK as one undo step (select.editSaved takes `edits`).
The menu stub count in the ROADMAP goes from 16 to 15.
A regression test for non-ASCII paths. It saves to a Japanese file name, with
full-width brackets and a middle dot, inside a Japanese folder. It checks the
document's title, reopens the file with its Japanese text, and exports SVG, PNG
and PDF to Japanese names.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Dragging a bounding-box handle of area type with the Selection tool now
resizes the type area: the text reflows at its own size (threads too)
instead of being scaled, and History says "Resize Type Area". With other
objects selected they scale as before while the area type resizes its
frame. Direct Selection picks the frame's corners and edges and reshapes
the area ("Reshape Type Area", new command text.reshapeArea), and the
selected frame's anchors and outline show. Area Type Options gains Width
and Height. Point type, rotation, Object > Transform > Scale, the Scale
tool and the Transform panel still transform the type.
Root cause: every handle drag previewed object.transform, which folds the
matrix into the text's xf (TextObject::transform), so area type scaled its
glyphs; and Direct Selection's anchor hit test only looked at paths, so the
frame's corners could not be picked. object.transform now takes
typeAreas, which reshapes the frame in text space (xf^-1 * m * xf), and
the tools use it and text.reshapeArea.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The tool chose scale, distort, perspective or shear from the modifiers at
the press only, but a Cmd press on the canvas is the temporary Selection
tool, and the documented gesture is to start dragging and then hold Cmd.
A handle drag now follows the modifiers held during the drag; when the
operation changes, the interaction restarts so the undo step is named
after what was done.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The export tool's description, the docs' open_file and export rows and the
CLI's run/convert usage listed only some formats. The CLI usage now prints
the writable extensions next to the readable ones, both generated from the
format table (export_extensions), the export tool points to its generated
format enum, and the docs name EPS, DXF, EMF, WMF, SVGZ and templates.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A type node's fill, stroke and strokeWidth now come from its first run, as
paint.proxies and its strokeOptions already did; fills and strokes of the
type object itself are reported apart as objectFill, objectStroke and
objectStrokeWidth.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
document.exportSelection (the MCP export tool's `selection: true`) now
replies with `format` and the encoder's `warnings`, the same list a
whole-document export gives: lossy WebP written lossless, blend modes left
out of SVG Tiny 1.2, PDF layers at 1.4 and the rest. A selection export
that approximates nothing replies with an empty list.
Root cause: export_selection encoded with `encode`, which drops the
warnings encode_with_warnings returns, and its reply carried only path,
bytes and bounds. The params doc and docs/mcp.md now describe the reply.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The Type tool passed `y + 12` to `text.create` for a dragged frame, and
`text.create` puts the frame's top-left at (x, y), so every frame landed
12 pt below the marquee the user dragged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
EPS files the reference app saves (its procsets in the prolog, a binary header with a TIFF
preview) now open as editable art instead of failing with "undefinedresource in Generic". When
their PostScript still can't be read, the palette preview they carry is placed with the usual
warning instead of the file being refused.
Root cause: five interpreter gaps and one preview format. The Generic and Category resource
categories didn't exist, so categories couldn't be made from Generic (now Category holds the
standard categories, defineresource in Category makes new ones, undefineresource is added);
clipsave/cliprestore were unknown (now a clip stack in the graphics state); systemdict's
languagelevel was the operator, not the integer 3; getinterval copied arrays and strings
instead of sharing their storage, so astore into a subarray was lost (arrays and strings are
now views into shared storage, and readstring/readline return the part they fill); and
resourceforall was unknown (now enumerates matching names with * ? \ templates). The preview
is an 8-bit palette TIFF with an alpha sample, which the image crate can't decode: a small
bounds-checked reader handles that layout (either byte order, any strips).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Strings in code stay English and are the lookup keys; `crates/ui-egui/src/i18n`
maps them to display text at render time from one `*.tsv` catalog per language
(`LANGUAGES` registry, `tl!`/`t`/`tr`/`tr_ctx`/`tr_id`/`tn`/`fmt`). Command ids,
menu paths, `ui.menu.list`, the control channel, the CLI, MCP, status messages
and undo labels never see translated text, so agents, scripts and the existing
tests are unaffected (under `cargo test` the UI is always English).
Wired: in-window menus and the macOS native menu (rebuilt on a language change),
dynamic labels (Undo X, Reset X, Last Effect: X as templates), the command
palette (matches English and translated text), dock tabs, panel tooltips, the
toolbar (category names now cut by characters, not bytes), the dialog frame, the
shared widgets, the Preferences dialog, the shortcut editor, the hint and status
bars, the Home screen, and the literals of every panel and dialog.
Preference `interfaceLanguage` (User Interface › Language): `auto` or a code.
`auto` follows `VECTORCRAFT_LOCALE`, `LC_ALL`/`LC_MESSAGES`/`LANG`/`LANGUAGE`,
the macOS preferred languages and, on Windows (which sets no `LANG`), the user
locale from `HKCU\Control Panel\International`. `zh-TW`/`zh-HK`/`zh-MO`/
`zh-Hant-*` resolve to `zh-hant`; Simplified locales stay English until a
`zh-hans` catalog exists.
`zh-hant.tsv` (~2,850 rows) is clean-room, in the vocabulary used in Taiwan;
terms shared with PhotoCraft reuse its wording, vector terms use everyday
vector-drawing vocabulary. Tests: catalog well-formedness (placeholders,
ellipses, command ids, no Simplified characters), locale resolution, and two
coverage tests that fail when a menu string, panel/tool/preference label,
dialog button, hint-bar fragment or `tl!` literal has no row in a language
marked `complete_menus`. `VECTORCRAFT_I18N_DUMP=file cargo test -p
vectorcraft-ui-egui dump_source_strings` lists the strings for translators.
Also: two clippy findings under rustc 1.97 fixed in passing.
Integrated with main (squashed onto it, 42 commits on): the Czech interface
(#213) moves from its Rust table to `i18n/cs.tsv` on the registry (all 556
rows, Czech plural rule, its menu-coverage and glyph tests kept); Find Font's
resolution notes and the canvas context menu's labels are translated and
covered by the catalog tests (`menus::CONTEXT_LABELS`); the glyph test checks
CJK text only when the craft-fonts build input is embedded (#220 moved the
Japanese fonts out of this repository); a language saved by an older version
carries over to `interfaceLanguage`; Windows' locale lookup runs reg.exe by its
full path.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
list_says_how_each_font_resolved expected Source Sans 3 Black to be a
substitute, which holds only while the bundled family is the only Source
Sans 3: on a machine with the full family installed (it ships with many
font packs), Black resolves exactly and the test failed. It now asks for
a style no Source Sans 3 has, so the substitute path is tested anywhere.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Conflicts with #214 (fonts by PostScript name): this branch's name index (FaceKeys,
aliases) stays, and #214's `by_postscript_name` (used by PDF import) reads a PostScript
map filled from it, keyed by the same normalized names.
Removed: scanning ~/Library/Application Support/Adobe (CoreSync/plugins/livetype and
Fonts), and the preference that switched it (Type › Include Fonts of Other Design Apps'
Folders). Fonts activated through another vendor's subscription are licensed for use
through that vendor's own sync; reading their files from VectorCraft would go around that
licence, and the project's clean-room rules keep VectorCraft out of other vendors' files.
Fonts the user has installed in the system or user font folders are still found by every
name, as before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
macOS keeps faces for its own interface under family names that start with
"." (".SF NS", ".LastResort", ".Aqua Kana"…); Mac apps don't list them. Since
the font menus list every installed family (#36), they showed 74 of them on a
stock macOS, at the top of every list.
- FontDb::menu_family_list: family_list without names that start with ".",
cached like it. Only a leading dot hides a family.
- Type › Font, the family dropdowns (Character, Properties), text.fontList and
Refresh Font List's count use it.
- Lookups by name still see every family: documents and fallbacks that name
a hidden family still find it (face, has_family, find_family, families,
text.fontList {family}, Fonts in Document's missing check).
- Test: tests_sysfonts::font_lists_leave_out_hidden_system_families_which_still_resolve
(a temporary font folder); the engine and UI font-list tests compare with
the listed families.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Choosing an Effect > Pathfinder item (Add, Intersect, Exclude, Subtract...)
with several loose objects selected now groups them and applies the live
effect to the group, in one undo step, so the result shows at once and stays
editable. A selection that already holds a group, or a single object, works
as before, and agents get `grouped: true` back when it happens.
Root cause: Pathfinder effects combine the contents of a group or layer.
Applied to loose objects, each one got an effect with nothing to combine, so
nothing visibly changed, while the Pathfinder panel's commands (found by the
command search) act on the selection directly.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The Live Paint Bucket now fills areas enclosed by any paths. Crossing
lines enclose faces it can fill, a line across a shape splits that shape's
face in two, and an area enclosed by paths becomes a face even when nothing
fills it (it starts unfilled). Edges are the paths cut wherever another path
meets them, keeping their stroke, so Shift-click paints each piece. Closed
shapes give the same faces as before.
Root cause: Live Paint built its faces from the Pathfinder regions, which
divide the filled areas of closed shapes. An open path counted only as the
area it would fill if closed, which for a straight line is nothing, so lines
divided nothing and the bucket found no face to fill. A new planar map in
vectorcraft-pathops (live_paint) sweeps every path as an edge: an open path
goes in traced there and back, with crossing counts that never cancel; faces
are walked on the resulting graph (holes included), and their fills and the
edges' sources are found geometrically. The sweep runs in an exactly sheared
frame, because it mishandles a horizontal line traced there and back.
Make, Merge, the bucket and Live Paint Selection all use it;
interior_point moves to vectorcraft-pathops with it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Builds on the vertical type from #145 (same data, commands and tools).
- Kinsoku (the strict set, horizontal Japanese too): no line starts with a
closing bracket, 、。, a middle dot, colon, !?, the long vowel mark,
iteration marks or small kana; no line ends with an opening bracket.
Hyphenation breaks obey it too.
- Tate-chu-yoko: a run of two or three half-width digits (10月, 第100回)
stands upright across the column as one block one em long, squeezed to
the em when wider; longer numbers stay on their side.
- Upright or on its side by Unicode's Vertical_Orientation: CJK and
symbols upright, marks that need a vertical form on their side when the
font has none; upright glyphs advance at least one em, centred, and turn
about the ideographic em box centre so they sit on the column's centre
line. Point type keeps its anchor on the first column's centre line.
- `vrt2` is no longer requested: its glyphs are already on their side, so
with the layout turning Latin and digits too they came out upside down
(and tate-chu-yoko blocks broke) in fonts that have it, e.g. Hiragino.
- The line→text mapping is one affine (`TextLayout::line_xf`, an exact
quarter turn plus the point-type offset); `logical_point` and
`physical_point` use it. Pixel snapping skips vertical type.
- Tests: kinsoku in both directions, tate-chu-yoko blocks, upright vs
sideways, anchor on the centre line, arrows along columns.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Documents made elsewhere often name a font differently from the family
name VectorCraft lists: a localized family name (ヒラギノ角ゴシック), a
legacy family/style pair (Hiragino Sans + W3) or a PostScript name
(HiraginoSans-W3). FontDb now indexes every name in a face's name table
and resolves a request to the right face, telling the caller whether it
matched exactly, by another name, by substitution or not at all
(FontMatch).
- text: FaceKeys/aliases, FontDb::resolve, canonical, find_postscript;
fix a race where a face loaded by another thread was reported missing.
- engine: text.fonts reports status, the resolved face and missing
glyphs; Replace Font resolves the same way.
- render: substituted-font highlighting uses resolve, so a font found
by another name is no longer shown as missing.
- Find Font notes how each font was found.
- Preferences › Type › Include Fonts of Other Design Apps' Folders
(macOS, on by default).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Typing with an input method (the macOS Japanese IME and the like) used to
insert the raw keys (romaji came in as ASCII): egui's Ime events weren't
handled.
- Tools gain `ime_preedit` / `composing` / `ime_caret`; the Type tool lays
the marked text out in place inside its typing session, underlined (the
clause being converted thicker), and commits it through `text_input`.
Character ranges from the IME convert to byte ranges in one place. A
cancelled composition leaves no empty undo step; an empty clear with
nothing marked never deletes the selection.
- The UI sends Text and Ime events to the tool in order, sends IMEOutput
every frame while text is edited (what lets egui-winit enable the IME
and place its candidate window at the caret, or the converting clause),
and holds back keys, Enter/Escape, clipboard, single-key shortcuts and
menu commands (Undo, Cut, Select All… also from the native menu) while
composing. A click, a tool switch or a style change ends the
composition with the marked text kept as typed; on macOS the IME is
told to discard it (NSTextInputContext.discardMarkedText) so it doesn't
come back in the next input.
- Undo during a typing session also ends the Type tool's session, so its
caret and marked text follow the document.
- Tests: tools (char → byte ranges, underlines and candidate position),
engine (compose → commit is one step, cancel leaves no step, a click
commits, Undo mid-composition is safe), ui-egui (synthetic Ime events:
compose, commit, ⌘Z held back, IME output, click and app switch).
Checked by hand on macOS 27 with the Japanese IME (romaji input, convert,
commit, cancel, Undo/Redo after commit, click away, Latin mode, app
switch). Not checked: Windows and Linux IMEs, the web build.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Brandon Thomas
bflatastic
EthanSuquet
Elias Witt
theironchef
imachus
Fahri Meral
Sarah Wesker
Hiroshi Inukai
WildfootW
midasdf