A PDF, and the PDF part of a PDF-compatible .ai, writes a shape filled
with a gradient as a clip followed by a shading fill: the shape, W n,
then sh, which fills the clip's bounds. The importer kept that as a
clip group of the shape (a clipping path) and a rectangle the size of
its bounds filled with the gradient. The Shape Builder skips clipping
paths, so it built its pieces from the rectangles: two overlapping
gradient-filled diamonds gave three rectangles. The Fill swatch showed
? for such a selection, since it held groups.
A clip whose only content is one rectangle covering the clip, filled
with a gradient placed on the page and nothing else, now comes in as
one path: the clip's outline, its subpaths closed, with the
rectangle's fill, opacity, blend mode and mask. It draws as before and
edits like a gradient-filled path drawn in the app. Clips holding
anything else (a flat colour, several objects, a gradient that doesn't
cover the clip, a shading that doesn't extend past its ends) stay clip
groups.
Tests: a diamond clipping a shading imports as one closed path of four
anchors with the linear gradient where it was, and a gradient covering
only part of the diamond or a flat colour stay clipped. Two overlapping
diamonds in a hand-written PDF, clicked on their overlap with the Shape
Builder, give what is left of each (9000 pt² each) and the overlap
diamond (3000 pt²), each with its gradient; main gave three rectangles
of 12000 pt².
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Add a Korean catalog (crates/ui-egui/src/i18n/ko.tsv, 4,151 rows) covering
every menu string, every tl! literal and the status and error messages, and
register `ko` in i18n::LANGUAGES (plural_none; every ko-* locale resolves).
- tests: korean_reads_as_korean, ko locale tags, ko in KEEPS_MENU_NAMES and
COMPLETE_MESSAGES; is_cjk covers Hangul so the bundled-font glyph check
leaves Hangul to installed fonts, as it does CJK without craft-fonts.
- Hangul is drawn with an installed system font on the desktop; the web
build still needs a Korean UI face in craft-fonts.
- docs: localization-parity (Korean row, 12-language score 54% -> 63%,
60-100 h), ROADMAP, README, development.md, gaps.md G16,
target-app-parity.
Clean-room: written from the meaning of the English strings in ordinary
Korean vocabulary; no other product's Korean localisation was used.
AI-drafted; native-speaker review welcome.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A CMYK page with transparency is drawn through an alpha soft mask that covers the page (a
non-isolated group). Its cover was black: opaque as alpha, but nothing shows in a reader that
takes the mask's luminosity. A white cover is the same mask either way.
Setting a tiling pattern went through every pattern made so far to
find one with the same PaintProc, cell and color, and went through the
document's patterns again to name a new one. Each spot ink was looked
up in the list of every ink used. When the reading ended, each ink and
each pattern became a swatch after a search of all the swatches, and
the second object of each clipping group of two was looked up in the
list of every shading painted. For a file that makes n of any of these,
the searches did work that grows with n squared: a dev-profile build of
vectorcraft-cli info took 99 user CPU seconds on a file that makes
160,000 tiling patterns, 35 on one that uses 160,000 spot inks, and 3.2
on one that paints 160,000 shadings through clips.
The patterns are now kept in a hash map, under a hash of the
PaintProc's storage, the cell, and the color, tint and swatch that an
uncolored pattern is set in. A lookup compares the patterns under the
same hash, with the comparison it made before. The inks are kept in a
map by name, and the shadings are kept in a set. The swatch names go
in a set while the swatches are made. The reader's document gets its
patterns only from tiling, so the name search always ended at Pattern
n+1 after n patterns, and that name is now made directly. A pattern
whose cell has a NaN corner never compared equal to another, and it is
no longer kept for lookups. The three files now read in 1.3, 0.6 and
0.9 user CPU seconds, and the documents read are the same as before.
Tests: a program that makes 200,000 tiling patterns and uses 200,001
spot inks reads within 20 seconds, with the patterns named in order;
a pattern made again from its prototype, and an uncolored pattern set
again in a color it was set in, are the same pattern.
The details name the copy's images, so the new document written before them paired with the old
details lost the images placed since the last copy (and a first copy lost all of them). Until the
new document is in, the details name the old and the new images.
Save serializes JSON straight into the compressor and writes images from their Arc into the
file; open reads each image into its own buffer, and NodeKind reads its tag first instead of
buffering every node. Already-compressed images are stored rather than deflated again in the
same gzip stream. Limits apply per section, so any file VectorCraft writes opens again. Data
Recovery writes each image once, content-addressed, instead of every interval. Native files open
on a background thread. Files stay readable by main both ways; JSON-only saves are identical.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VmMmc1J8XeNLxhRihWv9cc
Rarely set fields move into one optional boxed NodeExtras, and large NodeKind payloads (images,
live shapes, blends, envelopes) are boxed. Every object, and every undo step that copies it,
shrinks: a 50k-path move keeps ~24.5 MB per step instead of ~40 MB. Files stay byte-identical
(Node keeps its flat JSON through a borrowed Serialize and a from-wire Deserialize, no flatten),
and loading is slightly faster. Undo and redo reuse their step's size estimate instead of
walking the document again (8.3 -> 5.6 ms for a 50k-path move).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VmMmc1J8XeNLxhRihWv9cc
# Conflicts:
# crates/engine/src/cmd/graph.rs
# crates/svg/src/export.rs
Text geometry, clips, clip paints, effect art, brushes, live effects, pattern tiles and freeform
grids were capped by entry count and cleared wholesale on overflow, so a document past the cap
(e.g. more than 4,096 type objects) recomputed every entry on every frame. A FrameCache keeps
per-entry byte sizes and last-used frames, drops the least recently used entries of earlier
frames, and sweeps stale ones. Text geometry keeps its glyph outlines once instead of twice.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VmMmc1J8XeNLxhRihWv9cc
The navigator re-rendered the whole document on the UI thread on every revision, including
each drag step, and layer thumbnails re-rendered whole layers the same way. Both now keep the
last image while an interaction runs, render on a background worker once the document settles
(synchronously on wasm and in tests), and turn on image mips. Thumbnails are keyed by node id
with a Weak, not by a reusable address.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VmMmc1J8XeNLxhRihWv9cc
History was limited by step count only; a whole-selection move of 50k paths keeps ~40 MB per
step. Steps now carry a size estimate (only unshared subtrees are visited) and the oldest are
dropped beyond a byte budget, keeping the last 8. Documents entering history drop their id
index, selections are shared between steps, the command journal is bounded (held while an
Action records), and document resources are copy-on-write. Saved files are byte-identical.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VmMmc1J8XeNLxhRihWv9cc
Decoded images were kept per Renderer at full resolution and never evicted, so the canvas,
render worker, thumbnails and panels each held their own copy (96 MB per 24 MP photo).
They now live once in a process-wide LRU store keyed by content, with optional mip levels
(RenderOptions::image_mips, off by default so exports stay identical). Thumbnail renders
now sweep their caches, and shadow/effect rasters share a 512 MiB budget and match by Arc.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VmMmc1J8XeNLxhRihWv9cc
Simplify and Add Anchor Points edited the document whenever the
selection held a path, so a call that changed no path still recorded
an undo step, emptied Redo, marked the document modified and turned
the selected live shapes into plain paths. OK in the Simplify dialog
did the same. Simplify at its default options leaves a rectangle, a
star, a line and a path with one anchor as they are, and Add Anchor
Points leaves a path with one anchor as it is.
Both commands go through edit_paths, which now computes the new paths
before it edits the document. When no path changes, the command
returns without editing the document, and the reply
({ids, before, after}) has the same form and counts as before. When
some paths change, the command replaces only those paths and records
one undo step, so a path that Simplify leaves as it is keeps its live
shape.
One line in the evidence list of docs/target-app-parity.md names the
commands.
Disable, Enable, Unlink and Link Opacity Mask and Opacity Mask Options
(transparency.setOpacityMask) edited the document whenever a target
had an opacity mask, even when every mask already had the flags the
call set. Such a call recorded an undo step, emptied Redo and marked
the document modified.
When every mask a call acts on has those flags already, the call now
returns without editing the document, whether it acts on the
selection, on id or ids, or on the object whose mask is being edited.
The reply is still null, and the journal still records the call. A
call that changes a flag on any of the masks is still one undo step.
One line in the evidence list of docs/target-app-parity.md names the
commands.
setdash took a dash array of any length, and the graphics state kept
a copy of it. gsave, gstate, currentgstate and setgstate copied the
whole pattern with the state, currentdash made an array of it, and
each stroke and strokepath copied and scanned it. Each of these
counted as one operation against the operation budget. Under a
pattern of 1,000,000 entries, a pass of gsave grestore took 0.38 ms
of user CPU time, and a pass of strokepath on a line took 2.05 ms.
Loops of the two would reach the budget after about 1.6 and 2.8
hours.
setdash now raises limitcheck for a pattern of more than 65,535
entries, the most elements an array has in a typical PostScript
implementation (PostScript Language Reference, third edition,
Table B.1). Saved and copied graphics states share the pattern.
setdash and each stroke count each entry they read or copy as one
operation, and currentdash counts the array it makes against the
memory budget. Under a pattern of 65,535 entries, loops of setdash,
currentdash, stroke and strokepath now run out of the operation
budget, or the memory budget for currentdash, after 0.08 to 0.31
seconds of user CPU time.
Tests: a pattern of 65,535 entries is set, and a pattern of 65,536
entries stops the reading with limitcheck. Under a pattern of 65,535
entries, 4,000,000 passes of gsave grestore and 2,000,000 passes of
gstate currentgstate setgstate finish, and loops of setdash,
currentdash and strokepath run out of a budget. The test of these
loops reads each program on a thread of its own and fails when no
result arrives within 20 seconds.
These commands edited the document even when nothing changed, so the
call recorded an undo step, emptied Redo and marked the document
modified:
- Fill, Stroke, None, Apply Last Color and Apply Last Gradient with the
paint the objects have already, also on characters selected with the
Type tool.
- stroke.set and stroke.setAdvanced with the options the strokes have
already.
- paint.editGradient, paint.setGradientGeom and the freeform point
commands when the gradient comes out as it was.
- Invert and Complement on the Appearance panel's selected row when no
color changes.
- Swap Fill and Stroke and Default Fill and Stroke when every object
comes out as it was.
- appearance.setItem with the values the item has or an index no
object has, and transparency.set with an item and the values that
item has; appearance.removeItem with indices past every stack;
appearance.moveItem, Reduce to Basic Appearance, Clear Appearance,
the Eyedropper (appearance.copyFrom) and appearance.transfer when
every object comes out as it was.
These commands now return without editing the document. The replies
have the same form as before. On a graph's series or labels, a fill or
stroke that the art has already still records one undo step when the
graph does not keep that paint for them (for example after the
Eyedropper painted the art), and the graph then keeps it.
The Eyedropper test whose second copy changed nothing gives its target
a green fill first, which the copy turns red again.
One line in the evidence list of docs/target-app-parity.md names these
commands, apart from Invert and Complement.
The operation budget counted each operator once, however long a path
it made or read. arc and arct with a huge radius make up to about a
million curve segments, strokepath makes a piece for each dash, and
rectfill makes five path elements for each rectangle in an array.
pathbbox, clip, clippath, shfill, pathforall, gstate, currentgstate
and setgstate read or copy the whole current path or clip path, and
in an encoded user path a byte of the operator string can repeat the
next operator up to 223 times. In a loop, each pass counted a few
operations. vectorcraft-cli info was still reading each of 15 files
that run one of these in "loop" when "timeout 300" stopped it, after
28 to 128 seconds of user CPU time. With a radius of 50 instead of
2.9e30, the loop of arc ran out of the budget after 6.9 seconds of
user CPU time.
Each of these operators now counts the path elements it makes, reads
or copies as one operation each. Making or reading an element took up
to about as long as an operation in a loop of pop, and copying one
took less. strokepath also counts each entry of the dash pattern in
each period of the path. An encoded user path counts each entry of its
data array, each byte of its operator string and each operator it
runs. The 15 loops now end with "the program runs too long" after 0.06
to 2.35 seconds of user CPU time, and the loop of arc with a radius of
50 ends after 1.9 seconds of user CPU time.
Tests: a loop of each of these operators runs out of the operation
budget. The loops cover arc and arct with a huge radius, strokepath
with a fine dash and with dashes of length zero, rectfill with 2,000
rectangles, the eight operators that read or copy a path of 10,001
elements, and uappend on three encoded user paths. The tests read each
program on a thread of its own and fail when no result arrives within
20 seconds.
A shading's Function can be an array of functions. eval evaluated every
entry of an array and the arrays inside it, with only a depth limit. An
array of 100 entries that all refer to the same array one level down,
nested that way a few levels deep, is small in memory, but each
evaluation visited about 100^levels functions, and a gradient samples
such a function 33 times. vectorcraft-cli info on a file with four
levels of 100 shared entries had used 90 CPU seconds, with a 12 MiB
resident set, when "timeout 120" stopped it. eval also counted nothing
against the operation budget: a loop that paints an axial or radial
shading counted a few operations a pass, however many functions each
pass evaluated. On a file that paints a gradient of one Type 2 function
with N 2 in "{ sh shfill } loop", vectorcraft-cli info ran for
209 CPU seconds, and its resident set reached 4 GB before the program
ran out of the operation budget.
eval now counts the functions one evaluation visits, as linear_points
counts its nodes, and raises limitcheck past 257. The PostScript
Language Reference defines a shading's function as one function
dictionary or an array of one for each color component. The reader
allows 32 components, and the depth limit allows each entry a chain of
8 functions. An array with more entries than the visits left raises
limitcheck before it is copied. Each function visited counts 8
operations against the budget: one visit took as long as 6 to 10
operations, measured with loops of pop and of a name lookup.
Tests: a shading over four levels of 100 shared entries raises
limitcheck, and the square drawn before it stays; a shading over one
array of 256 entries is drawn as a gradient, and one over 257 entries
raises limitcheck; a program that paints 2000 shadings over the
256-entry array runs out of the operation budget. The test reads each
program but the last on a thread of its own and fails when no result
arrives within 20 seconds.
Select All, All on Active Artboard and Inverse select from selectable:
the visible, unlocked objects on visible, unlocked, non-template
layers and sublayers, or in isolation mode the visible, unlocked
objects in the isolated group or layer and in its visible, unlocked,
non-template sublayers. Other Select menu commands did not use it:
- Clipping Masks, All Text Objects, Stray Points and Open Paths
selected hidden and locked objects and the objects in hidden or
locked groups and sublayers.
- Point Text Objects, Area Text Objects, Brush Strokes, Bristle Brush
Strokes and the Same commands for symbols, fonts and text colors
checked an object's own flags and its layer's, but not those of the
groups and sublayers between them.
- Next Object Above and Below selected the object beside the selected
one even when it was hidden, locked or a sublayer.
- A saved selection picked from the Select menu selected again the
objects hidden or locked since it was saved.
- Direction Handles direct-selected the hidden and locked paths in a
selected group.
- In isolation mode, the commands in the first two bullets, the other
Same commands, All on Same Layers and a recalled saved selection
selected art outside the isolated group or layer. Next Object Above
and Below and Direction Handles did so only when the selection
already held something that was not inside the isolated group or
layer.
In isolation mode selectable checked the flags of the isolated group's
or layer's objects and sublayers, but not those of the group or layer
itself or of the groups and layers it is in. Select All, All on Active
Artboard and Inverse still selected the isolated group's or layer's
objects after it, or a group or layer it is in, was hidden or locked.
In isolation mode selectable now returns nothing while the isolated
group or layer, or a group or layer it is in, is hidden, locked or a
template layer. walk_selectable visits the objects selectable returns
and the objects inside them, leaving out hidden and locked objects
with what they hold. It visits nothing in a hidden, locked or template
layer, sublayer or group. The commands in the bullets above now select
from it. All on Same Layers takes the objects on the selected objects'
layers that selectable returns. Next Object Above and Below pass over
the objects walk_selectable leaves out, and leave the selection as it
is when none is left in that direction.
The params texts of Direction Handles and Recall Selection,
docs/mcp.md (Object selection, Saved selections) and one line in the
evidence list of docs/target-app-parity.md describe the change.
Overnight doc conflicts were resolved by keeping both sides, which left two or three copies of
some milestone and parity rows (M1–M3, M5–M7, M14; six area rows; one type row). Each is now one
row with every note. The 2026-10-12 progress row takes the 30 PRs landed since and the Film
Grain row that #1282 added separately.
A subpath that geometrically closes (e.g. an ellipse outline) was always emitted
with SPLINE group 70 = 8 (planar only). Autodesk's DXF reference defines bit 0
of group 70 as 'closed', and CAM software such as VCarve relies on it to treat
the cut as a pocket or profile rather than asking the operator to join endpoints.
Now `self.e.int(70, if sub.closed { 8 | 1 } else { 8 })`.
Test: extend lines_are_lwpolylines_curves_splines_and_fills_hatches to assert
the existing closed ellipse's SPLINE group 70 is '9' (1 closed + 8 planar).
Aaron Guo
BFlat
Johau91
skylar taylor-barrick
dotCooCoo
arhammusheer
webdevtodayjason
Royce