VariableKind::Graph binds a graph root; dataset rows supply CSV as graph.setData.
Make Graph Dynamic, Variables dialog, capture/apply, and pt-br for the new strings.
Co-authored-by: Cursor <cursoragent@cursor.com>
Load and save variable datasets as CSV for data-merge pipelines, and
translate the new Variables panel commands in Brazilian Portuguese.
Co-authored-by: Cursor <cursoragent@cursor.com>
# Conflicts:
# docs/target-app-parity.md
Store the registration point with the symbol and place instances so that point lands on (x, y). Symbol Options exposes the same 9-point control as Transform.
Co-authored-by: Cursor <cursoragent@cursor.com>
Expose existing Warp styles, orientation and percentage controls through the shared live effect dialog while preserving preview history and effect targets. Keep ordinary Enter owned by focused non-searchable combo controls.
Co-authored-by: Codex <noreply@openai.com>
assets.export of abutting fills must leave no ghost edge (#983 / #1037); panel thumbnails use the same Art Optimized supersample. Progress docs mark the gap closed.
Keep surviving drawable runs separate when Delete or Backspace removes an open-path anchor. Preserve appearance, selection and one-step Undo.
Co-authored-by: Codex <noreply@openai.com>
Make and Edit choose the Intertwine tool: hovering highlights the overlap
area under the pointer, a click swaps which object is on top there, and a
path drawn around overlaps (Shift: a rectangle) arranges every overlap
inside it; Esc or Enter goes back to the tool chosen before. The overlap
faces move to vectorcraft-tools, shared by the tool and the command.
Clicks inside a raised area pick the member drawn on top there, the
Object menu gets the Intertwine submenu, and the new labels and the
predicate message are translated in the complete catalogs. docs/gaps.md
records what landed under G13.
An Intertwine group keeps its objects as they are, in their own stacking
order; each member can be drawn over the others in an area of its own.
The group draws, hits, saves and exports through its live expansion, so the
members stay editable and Release gives them back unchanged.
object.intertwine.arrange picks an overlap area by a click point or by a
closed path drawn around overlaps and toggles which member is on top
there, or brings a member to the front or sends it to the back where three
or more objects overlap.
A gradient given to several objects fits each one's own bounds. With
span: "selection", a linear or radial gradient without its own vector is
placed once on the union of the targets' bounds, at its angle and aspect,
and every target gets that vector in document coordinates, so the ramp
runs on across them. A solid colour, a swatch, a freeform gradient or an
explicit start/end with it is refused before anything changes; the next
object drawn still fits the gradient to itself.
object.group, ungroup, the clipping mask and compound path Make/Release,
the Arrange commands, lock, hide, edit.duplicate, object.move, rotate,
scale, reflect, shear and the Pathfinder commands acted only on the
selection, so a script had to send select.set first. Given ids, they now
select those objects (select.set, with its checks) and run as on the
selection, in the same undo step; without ids nothing changes.
The Curvature tool built its curve as a uniform Catmull-Rom: both handles of a
smooth point a sixth of the chord between its neighbours, the ends of an open
path without handles. Uneven spacing overshot or looped, a J or U from three
points bent sharply at the middle with straight legs, and the end segments
looked flat.
- `curvature_spline` (draw2.rs) makes each stretch of three points follow the
circle through them: a smooth point's tangent is that circle's
(`smooth_handles`), each handle `chord / (3 cos²(α/2))` long, α between
handle and chord capped at 90° (`arc_handle`). The ends of an open path
mirror their neighbour's tangent in their chord (`end_handle`). A turn
tighter than 30° (`TIGHT_TURN`) shortens its handles, so a narrow V comes to
a point. Degenerate points become corners. `path.curvature` and the rubber
band use it.
- `catmull_rom` is unchanged for the Pencil and Paintbrush fitting.
- `bend_last` (from: smooth), `curvature_move` on a smooth point, the point
`curvature_insert` adds on a straight segment, and `path.convertAnchor` to
smooth without x/y (`curvature_smooth`) use the same rule.
- `curvature_extend` gives the new end its end handle unless Alt-clicked
(`corner` param on `path.curvatureEdit`).
- Neighbours of a point dragged on an existing path keep their tangents (a
Pen-drawn shape isn't reshaped); the path being drawn re-curves throughout.
Document Info › Objects counts the paths but didn't say how long they are,
which sign and print work reads there (an LED strip along the letters, a
cut line). `document.info` now adds up the length of every path it counts,
the pieces of compound paths included and guides left out, and returns it
as `pathLength` (pt); the Objects section shows it as Path Length in the
document's units, right after the Paths count, so Selection Only gives the
selected paths' length. The label is translated in the catalogs that
cover the menus; pt-br only takes strings the UI itself shows.
The Control bar's file name is a menu: Replace…, Go To Link, Edit Original,
Update Link and Placement Options…. Replace… (also in the Properties panel,
in place of Relink…) puts the picked file in the image's place fitted inside
the old bounds and centred, keeping the new file's aspect ratio instead of
stretching it into the old one. links.relink takes `preserve` (transforms,
bounds, fileDimensions, fit or fill) to override the Placement Options for
that read.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
With Limit to Library, a row's new colour is the library swatch itself: art
links to a global or spot colour (a Pantone-style spot colour stays a spot
colour in its own model, tints keep their tint) and a process swatch gives
its colour in its own model, instead of being converted to the model of the
colour it replaces. The swatches used join the document's swatches (one the
document already has, same name and colour, is reused); recolor.apply
reports them as `swatches`. A folder button next to Limit to Library loads a
library file (.ase, .acb, .gpl, .vcswatches or a document's swatches) and
limits to it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The reference app's Scribble Options have a Variation for each of Path
Overlap, Curviness and Spacing. Scribble had one `variation`, which
moved each line's ends both along the line and across it, and every
turn of a scribble took the same curviness. An angle such as 270 was
stored as given.
Scribble now takes `overlapVariation` (pt, default 0.5), which moves
the ends along the line, `spacingVariation` (pt, 0.5), which moves
them across it, and `curvinessVariation` (%, 0), which varies the
curviness from one turn to the next. The dialog shows each Variation
after the option it varies, and the Appearance panel lists all three.
A Scribble saved or scripted with the earlier `variation` reads it as
the Path Overlap and Spacing variations. With the defaults, and with
the earlier `variation`, Scribble draws the same outline as before.
`effect.apply` and `effect.setParams` store an angle above -180 and up
to 180, so 270 is stored as -90, and the dialog shows the stored value.
The Appearance panel's inline editor lists an effect's values in
their current form, as the dialog does. It listed an earlier key next
to the defaults of the keys that replace it.
Docs: docs/effects-parity.md (Stylize row), docs/target-app-parity.md.
Copy and Cut made the system clipboard's formats in the frame that ran
them. The formats include the SVG markup and a PNG of the selection at
1 px per pt, up to 4,096 px a side. In a release build on Windows, Copy
of 50,002 rectangles on a 1600 x 1200 pt page took 1.00 s in its frame.
The thread that writes background saves and exports now makes the
formats from a copy of the clipboard, and the app writes them in a
later frame. The same Copy takes 0.18 s in its frame, and the SVG and
PNG are the same bytes. While that thread has a save, an export, a print or
a recovery copy to finish, Copy and Cut make and write the formats at
once. A later Copy or Cut, or text copied in a field or with the Type
tool, drops formats that are not written yet. Until they are written,
Paste on the canvas takes the internal clipboard, and while a text
field or the Type tool has the keyboard the app waits for them.
Docs: docs/mcp.md, docs/target-app-parity.md.
Effect › Pathfinder left type in the group out of all ten operations,
on the canvas, in Expand Appearance and in exports. The text outliner
returns type as a group with one path per run. `leaves` in
crates/effects/src/group.rs took that group as one leaf, and
`geometry` gives no path for a group. A 100 x 100 pt square grouped
with a 100 pt "H" inside it kept all of its 10,000 pt² after Subtract
and Expand Appearance. In the shape modes a member group with type in
front also lost its paint, because it took the style of the outline
group, which has none.
`leaves` now takes the paths and compound paths of the outline art at
every depth, as it does for a member group, and gives each one the
type's opacity and blend mode. The same square keeps 8,684.88 pt² after
Subtract and Expand Appearance, its 10,000 pt² less the glyph's
1,315.12 pt².
Docs: docs/target-app-parity.md.
Object › Artboards › Fit to Artwork Bounds and Fit to Selected Art ran
artboard.fitToArt and artboard.fitToSelection without an index, so
they resized artboard 1 whichever artboard was active. The Artboards
panel menu's Fit to Artwork Bounds already passed the active artboard.
The app now passes its active artboard as the index of the two
commands when a call names none, as it passes the artboard to Align to
Artboard, All on Active Artboard and Paste in Place, in Front and in
Back. That covers the menu items, keyboard shortcuts, the command
palette and MCP's run_command in remote mode. A batch passes its
artboard as the index to their steps that name none, and a recorded
action leaves the index out, so an action resizes the artboard that is
active when it plays. The params texts of the two commands and of
command.batch describe where the index comes from.
Docs: docs/mcp.md, docs/target-app-parity.md.
For Auto Size area type in columns, every text edit, each keystroke of
the Type tool included, laid the text out in one column and then laid
the columns out at that height and at each step of a bisection for the
least height that holds the text. In the release-build timing test
auto_height_in_columns_typing_timing (stories of 5,000 to 50,000
characters at 9 pt in a frame 500 pt wide), a keystroke laid the
columns out 18 to 22 times, and typing at the start of the story of
50,000 characters in 2 columns took 1,670 ms a keystroke.
The bisection now runs on the lines of the one-column layout.
columns_hold computes where those lines would sit in the columns, with
each column's first line at its first baseline below the column's top
(LineInfo::first_baseline, new in vectorcraft-text). The columns are
then laid out at the upper height the bisection ends on and, when they
hold the text there, at the lower one, unless that is the 1 pt it starts
from. When they hold the text at the upper height and not at the lower
one, Auto Size takes the upper height. When taller columns always hold
the text, the old search ends between the same two heights. When the
text wraps around other objects or does not fit the one-column layout,
or when that check fails, the columns are laid out at every step as
before. The same keystroke now takes 292 ms.
Docs: docs/target-app-parity.md.
On each pointer move of an Alt-drag with the Perspective Selection
tool, Session::preview put the drag's snapshot back and ran
perspective.move with copy again, and persp_edit copied every selected
object again. An Alt-drag of a plane widget with the Perspective Grid
or Perspective Selection tool ran perspective.plane.move with objects
"copy" again, and plane_move copied every object on the plane again.
Both commands now take their copies from preview_copies, as
apply_transform does for the Selection and transform tools: the first
preview copies the objects into a clone of the snapshot, and each later
preview of the same objects on the same snapshot maps only those
copies. The copies keep their ids, and the document, the selection,
the undo step and the journal are the same as before. When the stored
grid still lists attachments of its own, as grids saved before
attachments were kept on the objects do, both commands copy on every
preview as before. With 25,000 rectangles attached to the left plane,
a release build of vectorcraft-cli took a median of 424.8 ms for each
Alt-drag move of the objects after the first preview and takes
328.9 ms with this change. An Alt-drag of the plane's widget took
340.3 ms a move and takes 256.9 ms.
Docs: docs/target-app-parity.md.
File › Save as Template always wrote a .vctemplate. The save panel
offered only that type, file.saveAsTemplate ignored its format param,
and a name typed with .ait got .vctemplate added. New from Template
already opened a .ait as an untitled document.
Save as Template now also writes a .ait: the PDF-compatible file Save
As writes for .ai, carrying the document. The desktop save panel lists
VectorCraft Template first and the .ait type beside it, and the web
build's Save as Template dialog gets a Format menu with the same two
types. file.saveAsTemplate writes .ait for format: "ait", or for a path
ending in .ait when no format is given, and takes the .ai options;
file.formatOptions answers for ait. .vctemplate stays the default.
Exports and Export As still do not write .ait.
Docs: docs/mcp.md, docs/file-format-parity.md (.ait row),
docs/target-app-parity.md.
Paste takes a file copied in a file manager by its flavor, which
file_flavour tells from the file's first 4,096 bytes through
fileio::detect. detect names gzip data SVGZ unless it is a compressed
native file, and file_flavour gave no flavor for the format id svgz.
So a copied .svgz file did not paste as its art, although
clipboard.importSvg, the SVG flavor's command, reads SVGZ data.
A copied file whose gzip data unpacks to SVG markup now pastes through
the SVG flavor. file_flavour unpacks the head it is given with the
gzip decoder the SVGZ reader uses, up to where the head ends, and
checks the first 4,096 unpacked bytes with looks_like_svg, as detect
checks an uncompressed file, reading them as UTF-16 after a UTF-16
byte order mark. Other gzip files get no flavor, as before.
Docs: docs/target-app-parity.md.
Paste takes a file copied in a file manager by its flavor, which
file_flavour tells from the file's first 4,096 bytes through
fileio::detect. detect names PDF or PostScript bytes in a file called
.ai or .ait by that extension, and file_flavour gave no flavor for
the format ids ai and ait. So a copied .ai or .ait file did not paste
as its art, while the same bytes in a .pdf or .eps file did.
A copied .ai or .ait file whose bytes are a PDF now pastes through the
PDF flavor (clipboard.importPdf), and one whose bytes are PostScript
pastes through the EPS flavor (clipboard.importEps). Other files
with those extensions paste as before. A .ai saved without its PDF
part also gets the PDF flavor, and Paste answers it with
clipboard.importPdf's error, as it does for the same file named .pdf.
Docs: docs/control-protocol.md, docs/development.md,
docs/target-app-parity.md.
document.pdfInfo and clipboard.importPdf, given a PostScript file (an
EPS, or an .ai in an older PostScript format), failed with "this is a
PostScript file (...), which can't be opened yet: save it as PDF,
PDF-compatible .ai or SVG". document.open and file.place read the same
file through the EPS reader. The error told users to convert a file that
VectorCraft opens.
The error now says the file is not a PDF and that File › Open and
File › Place read it as EPS. The 13 catalogs that carry the message get
the new text.
A PDF page's UserUnit (PDF 1.6) sets the size of its user space unit
in multiples of 1/72 inch. The PDF reader has applied it since #1894, so
a page with /UserUnit 10 and a 36 × 50.4 media box opens at 360 × 504
pt. A .ai whose editing data has its artboards in the page's own units
then no longer matched its PDF part. It opened from the PDF part,
without its layers, its hidden objects and the art outside its
artboards, and with its artboards in a row instead of where the
editing data puts them. document.pdfInfo gave such a page's size in
points and its boxes in the file's numbers.
Editing data whose first artboard has the page's size in the page's
units, and not in points, is now compared with the PDF part scaled
down by the UserUnit. The document read from it is then scaled up by
the UserUnit, with its stroke weights, dashes, effects and pattern
tiles. When the layers can't be read, the artboard layout taken from
the editing data is scaled the same way. document.pdfInfo gives the
boxes in points, as it gives the size.
Docs: docs/mcp.md, docs/target-app-parity.md.
When a .ai file, or an EPS with editing data, opened from its editing
data, the reader converted each embedded 8-bit CMYK image to RGB and
stored it as a PNG, so the document lost the image's ink amounts. When
the same .ai opens from its PDF part, the PDF importer keeps them in a
CMYK TIFF.
The reader now stores those samples with ImageBlob::cmyk_tiff, as the
PDF importer does (crates/pdf/src/import_image.rs). An alpha channel
that hides part of the image becomes the image's opacity mask, a
grayscale image over it, as in PDF import. Grayscale, RGB and 1-bit
images still open as RGBA PNGs.
Docs: docs/file-format-parity.md, docs/target-app-parity.md.
The reference app resets tracking and kerning to 0 with Cmd/Ctrl+Alt+Q.
VectorCraft had no command for it.
type.resetTracking (Reset Tracking and Kerning, default Cmd+Alt+Q) sets
tracking to 0 and kerning to Auto, which the Character panel shows as 0,
in the text the Type tool has selected, or else in all the text of
every selected type object. At a caret it resets only the kerning of
the character before it, where Alt+Left/Right kerns. It is one undo
step. The range logic of type.step moves into a shared type_ranges.
When a call gives ids, type.step and type.resetTracking change all the
text of those type objects even while the Type tool edits text, as
paint.setFill does. Before, type.step changed the Type tool's selected
text instead.
Docs: docs/mcp.md, docs/target-app-parity.md.
The Color panel and the Color Picker preview a colour in an interaction.
While typing, that interaction joined the open typing session and
previewed from the document before the typing, so new type came back as
its placeholder text. Typing in progress is now kept as its own step
before another interaction begins, as Undo already does.
object.setBounds read x, y, width and height with f64_or, which falls back
to the current value for anything that isn't a JSON number, so
{"width": "60 mm"} returned ok and changed nothing. object.move,
object.nudge, object.rotate, object.scale (sy) and object.shear did the
same with their optional numbers. A value that is neither a number nor
null is now a BadParams error naming the key, as a missing required
number already is.
The profile is registered in the process's colour registry only, so the
next `vectorcraft-cli run` (or the next launch of the app) doesn't know it
and edit.colorSettings falls back to the built-in profile. The command's
description now says so, and that the profile has to be loaded again in
each run that names it. Keeping profiles across runs would need a place in
the preferences to remember them, which is left for the maintainers.
document.export, document.serialize and document.exportSelection write
a PDF-compatible .ai (format: "ai"), and document.formats lists ai as
writable, but none of their params texts named it. The params text of
document.exportSelection named only png, jpg, webp, svg, svgz and pdf,
although the command writes all 20 writable formats. The description
of the MCP export tool and its row in docs/mcp.md named every format
of the tool's format enum except ai, and the range text and the row
said that only PDF writes several artboards.
The three params texts now list every format their command writes,
and document.export's range and useArtboards texts name .ai beside
PDF. The MCP export tool's description names ai, and its artboard and
range texts say that .ai, like PDF, writes one page per artboard. The
range text and the row also say that SVG, SVGZ and EPS write one file
per chosen artboard, and that rasters, DXF, EMF and WMF do so with
options.useArtboards: true.
Docs: docs/mcp.md (the export tool).
The document.exportForScreens params text listed png, png8, jpg, webp,
gif, svg, svgz and pdf for a row's format, and its settings keys left
out svgz. A row of another format VectorCraft writes, such as eps, was
refused with "Export for Screens writes png, png8, jpg, webp, gif, svg,
svgz or pdf". Rows of tiff, bmp, tga and psd are written too, and the
Export for Screens dialog offers TIFF rows.
The params text, its settings keys and the error now list the 12
formats a row writes: png, png8, jpg, webp, gif, tiff, bmp, tga, psd,
svg, svgz and pdf. Which formats a row accepts does not change.
Type is drawn as its glyph outlines under a geometry effect, in EMF and
WMF export, in DXF export when it is written as outlines, through
Flatten Transparency with Convert All Text to Outlines, and inside an
envelope. Those outlines differed from the type the canvas draws in
four ways. They had no underline or strikethrough bars. A character
stroke on scaled type kept its unscaled weight. An unplaced gradient was
fitted to the glyphs it painted instead of the layout box. A run's
pattern tiles stayed in document space instead of scaling and turning
with the type. A group's or layer's own fills and strokes also painted
the glyphs of the type in it without the bars.
The outlines now include each run's underline and strikethrough bars,
as the canvas has drawn them since #850. Each run's path is made in
text space with its unplaced gradients fixed to the layout box, then
placed by the type's transform, which also scales its stroke weight
and dashes and transforms its pattern tiles. The type's own fills and
strokes have their unplaced gradients fixed to the bounding box of the
layout box in the document, where the canvas fits them.
Geometry effects that move or scale the outlines move and scale the
type's gradients with them: the type's own effects, also inside an
envelope and after Flatten Transparency has converted the type to
outlines, and the effects of a group, symbol instance or envelope that
holds the type. outline_text marks its group with a new node field,
type_outlines, which is saved with the document. reshape_leaves maps
the placed gradients in each marked group through the map that takes
the group's bounds before the effects onto its bounds after them. A
scale out of proportion stretches a radial gradient into an ellipse
and turns the isolines of a diagonal linear gradient, as it does on
type scaled by its own transform. Placed gradients outside those
groups stay where they are.
Export Selection, Asset Export, the Assets tab of Export for Screens
and the rasterized art of CSS Properties cropped each object to its
visual bounds, which leave out live effects. A path with an 8 pt drop
shadow was exported with the shadow cut off at the path's edge, and
a group's own shadow was cut off the same way.
The crop is now the bounds the renderer measures for what it draws
(`vectorcraft_render::effect_bounds`): `painted_bounds` without its
room for brush art. A brushed path is cropped to its visual bounds as
before, and a Shaper group leaves out its transparent source art as
its visual bounds do. `assets.list` reports the same crop. The
renderer measures a shadow's or glow's reach on every side of the
object, so the image also has transparent room on the side away from
an offset shadow.
Docs: docs/mcp.md, docs/target-app-parity.md.
PDF, EPS, EMF/WMF and raster exports, and SVG exports that outline or
embed fonts, warn when type is set in a font family that isn't
available, but file.print and print.preview gave no such warning. A job
in a missing family printed in Source Sans 3 with no warning, and the
Print dialog's Summary showed none.
Both commands now add the exports' warning for the type that prints:
visible type that fits its frame or path, on the layers that Print
Layers picks, where what it paints (its effects applied) lands on the
paper of an artboard or tile that prints, inside its clipping paths.
The type in the symbols, inline graphics, patterns, scatter brushes,
opacity masks and placed documents of the art that prints counts too.
file.print reads the fonts before raster effects and Print as Bitmap
turn type into images, for PDF and PostScript output.
Docs: docs/mcp.md (Print), docs/target-app-parity.md (Fonts).
dataset.select applied a row's values in the order the variables were
defined, and it skips an object inside a hidden group. When a row
showed a group and set an object inside it, and the object's variable
was defined first, the object was skipped while the group was still
hidden: one selection applied the row in part, and selecting the row
again applied the rest. A text variable on type inside the group was
skipped the same way, and a row that hid a group skipped the objects
inside it whose variables were defined after the group's.
dataset.select, dataset.next and dataset.prev now apply the visibility
values that show objects first, outermost object first, then the other
values in definition order, then the visibility values that hide
objects, innermost object first. Of two variables of the same kind on
one object, the one defined last still wins.
Docs: docs/mcp.md.
Binding art in a hidden group or layer failed with "node N is locked",
even when nothing was locked. The error now says whether the object is
locked, is in a locked group or layer, or is in a hidden group or layer.
Make Text Dynamic and Make Visibility Dynamic reported their errors under
`variable.makeDynamic`, which is not a command id. They now name
`variable.makeTextDynamic` or `variable.makeVisibilityDynamic`.
A data set can hold values for only some variables: `dataset.new` and
`dataset.set` store only the values they are given, and selecting the
data set leaves the other variables alone. Update Data Set replaced the
active data set's values with the values the art reports for every
variable. After one Update, a data set that set only a visibility
variable also held the current text of the bound type, and selecting it
again wrote that text back. Update also dropped a variable's value when
its objects were locked or inside a hidden or locked layer or group,
because such objects report no value.
`dataset.update` now writes the art's current value for each variable
the active data set holds, and keeps the data set's value when the art
reports none. A variable that the data set does not hold stays out of
it, including one defined after the data set was captured;
`dataset.update {all: true}` also adds the other variables that the art
reports a value for. The reply adds `left`, the variables it did not
write. The Variables panel's Update Data Set button and menu item run
it without `all`.
Docs: docs/mcp.md, docs/target-app-parity.md.
The Pen's pointer showed the new-path asterisk before the first anchor
and also while a path was being drawn, where a click adds the path's
next anchor. A new Cursor::PenDrawing, the pen nib without a badge, now
shows there. The close, convert and join badges still show over the
first anchor, the last anchor and another open path's end. The asterisk
shows only where a click starts a new path.
The Continue, Add and Delete badges now also show after another path
was selected while the Pen was still drawing (for example with a
Cmd-click). A click there already continued the path or added or
deleted an anchor, but the pointer showed the asterisk.
View › Proof Setup had two color blindness items, protanopia and
deuteranopia. A third item, Color blindness – Tritanopia-type, proofs
the canvas with a tritanopia simulation, and the Separations Preview
panel's Proof Setup list offers it too. `view.proofSetup` takes
`target: "tritanopia"`.
The simulation follows Brettel, Viénot and Mollon (1997): a color
keeps its L and M cone responses, and its S response is replaced so
that it lies on one of two half-planes. Both contain the neutral
axis; one passes through the 485 nm color and the other through the
660 nm color. The neutral axis runs to the display's white, so grays
keep their value. The matrices are computed at
compile time from the published inputs cited in the code. The
protanopia and deuteranopia matrices are unchanged.
