A press on one part of a dimension now also checks its other parts for
a geometry effect, so an effect on an extension line keeps the drag
from moving the dimension (and from drawing the parts again without
the effect). In isolation mode, only art inside the isolated object is
compared with the dimension when the press decides between them.
With no points or lines picked, a press within 8 pixels of what a
visible, unlocked dimension draws (its lines, arrowheads and label),
where it is the topmost object that can be edited, followed by a drag
moves that dimension as one undo step. A segment of the art as near
the press as the dimension's lines and label, or nearer, takes the
press instead, and a click there without a drag still picks a point.
The part grabbed follows the pointer: a linear dimension's line moves
as far across as the pointer moves, an angular arc moves as far as the
pointer moves away from or toward the vertex, and a radius or diameter
line turns about the center as the pointer turns, or toward the
pointer when grabbed at the center. The dimension keeps what it
measures and its link to the art. With Smart Guides on, a linear
dimension's line lands on the level of a parallel dimension line, as
when it is placed, and of two lines on one level the one nearer the
pointer along it wins. Isolation mode, clipping masks and locked artboards
apply as they do to a click; with a part of a dimension isolated, the
dimension is not dragged. A dimension that a geometry effect on it, on
one of its parts or on a group around it draws elsewhere is not
dragged, except in Outline view and on a layer whose Preview is off.
dimension.update takes `at: [x,y]`, which puts a linear dimension's
line through the point, gives an angular arc a radius of the point's
distance from the vertex (at least 12 pt), or turns a radius or
diameter line toward it. With Real-time Drawing off, a dimension.update
preview that names its dimensions outlines only those dimensions.
A dimension pasted into a document that had none now marks the
document as holding dimensions (Document::insert), so the Dimension
tool drags it like one drawn there and edits refresh it. Opening a
native file sets the flag from the objects it holds, which covers files
saved after such a paste.
Docs: docs/dimensions.md, docs/target-app-parity.md.
A linear dimension line placed near another dimension's line lands on its
level only when that line is parallel. Before, the check compared the two
directions only, so a long line at a slight angle passed it: with its first
end within the snapping tolerance, the new line landed on a level the other
line had already left, and the guide ran at an angle between the lines. A
line is now parallel when it moves off its level by no more than 0.000001 pt
across the span of both lines.
A level that would put the line more than 1,000,000 pt from its points is
passed over. Before, the tool took it, the preview failed validation and
was not sent, and a click committed the earlier preview position. The next
level within reach is taken instead, or the line goes where the pointer is.
The level guide also needs Smart Guides on when it is drawn, so turning
Smart Guides off before the next pointer move hides it.
The nonparallel case in the placement test now puts the pointer within
reach of the first line's end, so it fails without the parallel check, and
the guide test expects exactly one guide.
colored_revolve_faces_leave_no_alpha_holes_at_one_to_one compared each
representation only with its own 3x render, so a 60% opacity lost or
applied twice by SVG or PDF import passed at both scales. The test now
also reads the alpha inside the body: 153 within 3 levels with a fill
opacity of 60%, 214 within 3 levels with 60% gradient stops (front and
back faces at 60% each), and 255 when opaque.
docs/target-app-parity.md gets a line for the file.write report on a
File object's length, hidden and readonly and for the test that lists
every name the script host adds, with the tests that cover them. The
test's comment on vc no longer names another app.
Setting a File object's length truncates or pads the file, and setting
readonly or hidden changes it, but script.check reported no outside
category for those assignments. It now reports file.write and
filesWritten for them: plain and compound assignment, ++ and -- before
or after, and the property name as a string literal in brackets.
A new integration test lists every global and member the script host
adds to the JavaScript engine (the own properties of the global object,
of each object or function it holds and of each function's prototype,
compared with a plain engine) and fails when one has no row. Each name
that reaches outside the document has a probe script, and the test
checks that script.check gives it the expected outside id. vc is listed
as not checked: script.check reads scripts written for the reference
app's scripting, and what vc.run does depends on the command. The test
also checks that the checker lists every preprocessor directive the
host reads, through the new DirectiveName::ALL.
File object properties:
https://extendscript.docsforadobe.dev/file-system-access/file-object/
object.align takes to: "auto", which aligns to the key object, key
anchor or key guide when the selection has one, else to the artboard
when exactly one object is selected and no anchor points are, else to
the selection.
New Action and Action Options have Choose Align To at playback
(automaticAlignTo on ui.action.new and ui.action.options), off by
default. With it on, playback sends to: "auto" for each object.align
step, without any artboard index the step stores, so the step uses the
active artboard; the stored steps keep their recorded params. Save
Actions writes the option, and a set that uses it is saved as file
version 7.
The dialog's strings have rows in every catalog, and docs/mcp.md
describes to: "auto", automaticAlignTo and file version 7.
Fixes#6972. Supersedes #7635.
Co-authored-by: yoriswind <yoriswind@yahoo.com>
While a linear dimension is placed, each preview starts again from the
document as the placement began. The tool now reads the lines of the
other dimensions from that document (the new ToolContext field
interaction_doc), not from the document a command changed in between, so
a dimension that a command moves or hides and that the next preview puts
back does not leave the placed dimension on a level the committed
document does not have.
The engine test now drags a dimension out, runs dimension.update or
object.hide on the other dimension, releases, and checks the committed
line against that dimension as the document has it. The tool test that
checks which dimensions read lines also covers radius and diameter
dimensions.
A click on the last anchor retracts its handle, an Alt click over the path
goes to the Anchor Point tool, and a join with an end that lies on the last
anchor merges the two ends. None of these adds a segment, so no rubber band
shows there. The measurement label measures to the anchor the band ends on
instead of the snapped pointer. A press and a hover no longer search for an
end at the snapped point when snapping left the pointer where it was.
Saved and recorded dimension options carried "arrowHead": null even
without a stroke head, so a recorded dimension.create failed in versions
that do not know the option. The field is now left out when it is null.
dimension.create, dimension.update and Dimension Tool Options still take
it. The name is read without a copy of its own, and a name longer than
any built-in one is refused before it is parsed. Through ui.dialog.set,
an arrowStyle set in Dimension Tool Options clears arrowHead, and a
built-in arrowHead sets the arrowStyle of its Arrow Style entry, as that
menu does. The Arrow Style field reads only a built-in head's name from
a draft.
The Stroke panel and Arrow Style look up a built-in arrowhead's name in
the catalog's arrowhead rows first, then its plain row. Every complete
catalog gets an arrowhead row for Feather, so the Feather arrowhead is
named as an arrowhead in each, and the de, gu, it, ja, ko, uk, vi,
zh-hans and zh-hant catalogs no longer show it in English.
Versions that have the stroke arrowhead but do not read arrowHead keep a
dimension's saved arrows until they draw the dimension again, and then
draw arrowStyle; versions without the stroke arrowhead do not open the
file. The params text, the doc comments, docs/dimensions.md, docs/mcp.md
and docs/target-app-parity.md say so.
The French Line Arrow row moves next to the other rows of its block.
The engine test that runs commands while a linear dimension is placed now
also locks the other dimension, as well as hiding and deleting it, and
checks that no guide runs along its level.
docs/dimensions.md says that layers and groups count toward the 20,000
objects the tool reads for levels.
Arrow Style in Dimension Tool Options and in the Properties panel lists
Filled, Open, Tick and None, then the other 38 built-in Stroke panel
arrowheads, each with its drawn preview. The dimension line draws the
chosen head at both ends.
A dimension's options take a new arrowHead value: null, or a built-in
stroke arrowhead name as stroke.set takes it. The menu writes arrowStyle
filled with it, so versions that do not read arrowHead open such a
dimension with Filled arrows. Documents and tool settings saved before
open unchanged. Custom arrowheads and unknown names are refused.
Fixes#8049Fixes#3222
While a linear dimension is placed with Smart Guides on, the tool reads
the lines of the other dimensions again when the document changes, so a
dimension a command hides, locks or deletes stops being a level, and one
shown again or grouped is a level again. The preview of the dimension
being placed is left out of the lines and out of the 20,000 objects the
tool reads.
Over the path's first anchor, a click closes the path on that anchor; over
an end of another open path, it joins the two, and the segment arrives
along that end's own handle. The preview ended at the pointer and arrived
without a handle, so the segment moved by up to the pick tolerance at the
click and changed shape when the other end was smooth. The preview now
ends on the anchor the click uses, with that anchor's handle, and picks
the anchor at the pointer before snapping, in the order a press does.
Part of #8296.
The level a new dimension line lands on now comes from the dimension
lines in the document before the dimension's first preview, so its own
preview is never a level. A dimension drawn earlier between the same two
points now acts as a level as well; before, every dimension of those
points was skipped.
- Of two lines on one level, a line the pointer lies across wins over a
line whose end is nearer the pointer.
- A dropped press while placing removes the level guide with the
preview.
- The reach uses the Smart Guides snapping tolerance helper.
- docs/dimensions.md says the rule applies to a dimension anywhere in
the document.
- The tests are split by behavior, with new tests for the same points,
pointer moves in small steps, the guide after a dropped press and
with Alignment Guides off, and the tie-break.
The engine test now names the head each style draws (Filled a filled
triangle, Open an open triangle, Tick a slash, None no head) instead of
comparing the dimension line with ArrowStyle::arrowhead, which the
engine itself calls. It moves to crates/engine/tests/it/dimensions.rs.
The Arrow Style previews take each style from a constant list in the
order of the dropdown entries instead of parsing the entry's value
every frame, and a test checks that each entry's value matches its
style.
Part of #8049.
A linear dimension placed next to another one had its line exactly where
the pointer was. Chained dimensions only lined up when the pointer hit
the other line's level exactly, which at a normal zoom it rarely does.
- With Smart Guides on, a linear dimension placed with its line within
the Smart Guides snapping tolerance of the level of a parallel line of
a visible, unlocked linear dimension already drawn has its line on
that level. This applies to two clicked points and to a segment
dragged out.
- With Alignment Guides on, a guide runs along that level across both
lines while the dimension is placed.
- Smart Guides off, a level further off, a locked dimension or one that
is not parallel leaves the line where the pointer puts it.
Docs: docs/dimensions.md, docs/target-app-parity.md.
A PDF/X-3 or PDF/X-4 export tagged every CMYK color with an ICC-based
space of the same profile that the output intent embeds. In PDF/X-3 and PDF/X-4, CMYK colors in the
profile of a CMYK output intent are now DeviceCMYK, and the profile is
written once, as the output intent's DestOutputProfile (ISO 32000-1,
14.11.5:
https://opensource.adobe.com/dc-acrobat-sdk-docs/pdfstandards/PDF32000_2008.pdf#page=641).
- With output.profiles none and a CMYK output intent, PDF/X-3 and PDF/X-4
write every CMYK color as DeviceCMYK, unless colors are converted to a
CMYK destination of another profile, which keeps its tag. Before, those
standards tagged every color whatever profiles said.
- RGB and gray colors stay tagged, and so do CMYK colors and images in
another profile.
- A CMYK image that carries the output intent's profile is DeviceCMYK,
and so is every CMYK image when none makes CMYK colors DeviceCMYK. A
CMYK JPEG then loses its ICC_PROFILE segments.
- Transparency groups blend in DeviceCMYK when CMYK colors are
DeviceCMYK, also when the file has no ICC-based CMYK space.
- Files without a standard keep the tags that profiles asks for.
- The Save PDF dialog's note for PDF/X-3 and PDF/X-4 says that CMYK
colors in the output intent's profile are written as DeviceCMYK.
Docs: docs/mcp.md, docs/target-app-parity.md and the document.exportPdf
params text.
Clear with `ids`, which Cut uses for the objects it copied, removed each
id from the selection with its own scan of the selected objects. It now
calls Selection::remove_all once, which gives the same selection.
With 50,000 objects selected and all of them cut, the selection step
takes 5.5 ms instead of 572 ms (release build, median of 7 wall-clock
runs, measured while other builds were running). The new tests compare
the result with removing each id in turn, including Undo and Redo, and
an ignored benchmark times both.
A Linear point picked near the end of another dimension's line landed
beside it when zoomed in: the tool's own pick skipped dimension artwork,
so that end pulled the point only within the Smart Guides tolerance,
while an anchor of the art pulls it from 8 pixels.
- With Smart Guides or Snap to Point on, a point that no anchor of the
art within 8 pixels takes lands on the end of a visible, unlocked
dimension's dimension line, or on the far end of its extension line,
within 8 pixels, with no link to that dimension. The end of an
extension line next to the point it measures is not an end, so an
anchor that a dimension already measures still takes the point.
- A point that nothing else places lands on the closest point of such
a line within 8 pixels, as it does on a segment of the art.
- With Smart Guides and Anchor/Path Labels on, a point on an anchor or
a line end shows the "anchor" label, and a point on a segment or line
shows the "path" label. A point on a line end is marked as an anchor.
- A press on a line end that moves a few pixels picks it, as a press on
an anchor does.
Docs: docs/dimensions.md, docs/target-app-parity.md.
A Shift-drag marquee sends every object it reaches as one select.toggle
{ids}. Each id used to scan and rebuild the selected objects, so K ids
against S selected objects took O(K*(S+K)). Selection::toggle_all and
Selection::add_all now handle the ids in one pass with a map of each id's
current slot, and the commands use them. The result is the same as
toggling or adding each id in order: an id toggled twice is selected
again at the end, and a removal clears that id's anchors, segments, key
and key anchor.
Select > Inverse looks each selectable object up in a set of the
selected ones instead of scanning the selection for each.
Selection::toggle_target, set_slices and set_guides use the same set
lookups.
crates/engine/examples/selection_bench.rs times the old per-id loops
against the new ones in one process. With 50,000 objects, toggling every
object off took 528 to 655 ms before and 2.5 to 4.6 ms after in 7 runs
(wall clock on a machine that was running other builds).
Under colored lights, a solid-color Inflate surface drew each facet over
its underlay as a fill and then a same-color round seam stroke. At a
partly covered edge pixel the two coverages add up past the facet's real
coverage, so a turned Inflate's near facets bled into the facets behind
them by up to 5 levels against the same art drawn at 3x and averaged down.
- lighting::seamed_fill turns a facet that is one closed polygon with an
area and no crossing edges into one path: the facet and the outer
outline of its round seam (lighting::footprint). solid_surface fills
that path once per channel. Other facets keep a fill and a seam stroke.
The 2 pt underlay, the surface clip and the channel groups are as
before.
- three_d_light_color::turned_colored_inflate_facets_and_seams_cover_their_edges_once
checks Inflate at X 50, Y 35 in an opaque fill: at most one channel
sample more than 3 levels off, by at most 4, and no alpha off.
- docs/three-d.md (Seams) and docs/target-app-parity.md describe it.
Under a colored lightColor or a lights list, Revolve drew each face of a
solid-color or gradient surface on its own, so where two faces'
antialiased edges met, pixels inside the surface let the background
through at 1:1.
- The faces now lie over an underlay of themselves under one clip to
their outline, as colored Inflate's facets do (lighting::sealed, split
into lighting::underlay and sealed). The surface's opacity and blending
mode apply once, to the sealed surface.
- In a group, each run of one member's faces in the shared depth order
takes its own underlay; a translucent member's run keeps its opacity,
blending mode and mask on the sealed run.
- A face with a part of a mapped graphic gives its face to the underlay
and keeps the part right after it.
- Faces in a gradient with transparent stops carry its opacity mask and
are drawn as before. Revolve (Classic) is unchanged.
- The paint-copy budget counts the underlay in validation and drawing.
Docs: docs/three-d.md (Seams), docs/target-app-parity.md.
With Smart Guides on, the Dimension tool's second point snapped onto the
construction guides through the first point (Smart Guides › Construction
Guides), so a click beside a segment straight below a first point that
was not on an anchor stayed on the vertical guide instead of landing on
the segment.
The second point now has no construction guides through a first point
that is not on an anchor. After a first point picked on an anchor, the
construction guides through it still apply. The second point still snaps
to anchors, centers and alignment lines, and with Snap to Perpendicular
or Snap to Tangent on it lands where the line from the first point meets
a segment square or grazes a curve.
Docs: docs/dimensions.md, docs/target-app-parity.md.
An unlinked aligned dimension that was reflected on its own kept its
offset, which is measured along the left normal of a to b. A reflection
turns that normal to the other side, so a line 30 pt above two level
points stayed 30 pt above them after a flip top to bottom instead of
going below them. The offset now changes sign when the transform
reverses orientation, as a reflection or a scale by a negative factor
does. Other transforms leave it as before, and linked, horizontal and
vertical dimensions keep their own rules.
Docs: docs/dimensions.md.
Under a colored light, Inflate's facets were antialiased one by one.
Where two facets met, their edge pixels added up to less than full
coverage. The solid-color composite then showed its black channel
backdrop there, so internal edges drew darker at 1:1, and the per-facet
gradient art let the background through.
- The solid-color composite draws, in each channel, an underlay of the
facets with 2 pt round seams between the black backdrop and the
facets themselves.
- The per-facet routes put each run of consecutive faces of one surface
over an underlay of its own faces, clipped to the union of the faces
and their seams. Runs that carry a gradient's opacity mask are sealed
with that mask. A face clip's footprint keeps only the outer contours
of its seam stroke.
- The paint-copy allowance for cut faces counts 12 copies per facet.
At 1:1, inside the surface each color channel of an opaque fill is
within 3 levels of the same art drawn at 3x and averaged down, and
every pixel is opaque. Before, 2,267 channels differed by up to 10
levels on the solid Inflate, and 1,088 pixels of the gradient Inflate
were below full alpha. Under white light, the rendered pixels, the SVG
text, the PDF read back and the expanded art of Inflate, Extrude,
Rotate and Revolve are the same as before.
Docs: docs/three-d.md (Seams), docs/target-app-parity.md.
Under a light that is not white, an opaque Inflate fill or stroke in a
solid color was drawn as one composite of all its facets, so the
object's solids were drawn whole, one after another: the stroke's faces
covered every face of the fill, nearer or not.
When an object's solids share one order (all opaque with Normal
blending, or one member's under its opacity mask), each facet of such a
solid is now drawn on its own, with the composite's paint, channel
light ramps and seam width, and takes its place in that order as the
faces of a gradient already do. A lone solid, and the solids of an
object with a translucent or blended fill or stroke, keep the one
composite.
A face drawn on its own is about 16 nodes, against about 6 per facet
in the composite of a whole surface. These faces go in one order while
they number at most 30,720 together, counted again after the cuts.
Past that, each such solid is drawn as one composite.
Docs: docs/three-d.md, docs/target-app-parity.md.
vectorcraft_testkit::temp_dir returns a bare path, so every engine test
that wrote into one left the folder in the system temporary folder.
- testkit: TempDir, a test's own folder made through temp_dir (same
private creation and ownership checks) under a tag no other TempDir in
the process uses, removed with its contents when dropped, also when
the test panics.
- Engine unit and integration tests hold a TempDir for the whole test
instead of calling temp_dir: graph, fileio, compare, journal, library,
shortcut set, spelling, store, web export, cancel, containers, build,
color settings, command sweep, font folder, removed font, model-based
and the font search fuzz.
- The DCS and swatch library tests' own helpers and the PDF editing
tests' shared vc-pdfedit-<pid> folder use a TempDir per test too.
- The font and link file search, store, color settings and font folder
helpers no longer delete the private folder temp_dir checked and make
it again with default permissions.
Clear, path.deleteAnchors, Cut, Copy and Cut Path scanned every selected
anchor or segment of a path once for each of its subpaths, so a path with
thousands of subpaths took time that grew with the square of their count.
The selection sets are ordered by subpath, so a new helper, subpath_refs,
reads one subpath's entries with a range lookup instead. The results are
unchanged.
With one selected interior anchor on each of 8,000 open subpaths of one
path, Clear took a median of 74 ms of thread CPU time before and 3 ms
after (release build, five runs each; the timings are noisy because the
CPU was shared and loaded).
Tests in delete_path_pieces.rs check exact surviving subpaths and their
order for anchor and segment selections across open and closed subpaths,
an untouched subpath, references to missing subpaths and indexes, one
Undo/Redo step and a second Clear, and Cut Path on several subpaths.