main's gate fails french_catalog_uses_nonbreaking_prose_punctuation on
the "{pos} of {total}: {name}" row of fr.tsv, added by #8017, which has a
plain space before the colon. The row now uses U+00A0 there.
Open #8053 fixes three other rows that main already has right, so it
does not clear this failure.
Saving a legacy .ai file numbered a gradient definition whose name was
taken by searching every definition written so far for each number it
tried, so a document with many distinct gradients without a swatch name
spent time cubic in their count naming them. Spot color names were
numbered the same way.
Both now keep the names given in a set and, for each name, the number to
try next. Names are never given back, so every number below that counter
is still taken, and the file is the same as before: the first free
"Name 2", "Name 3" and so on, with equal gradients sharing a definition.
- New tests compare the counters with a search of the names given over
3,000 names, and the whole written file both ways for colliding
gradient and spot color names.
- An ignored timing test writes 1,504 gradient definitions: in release on
Linux, the median of 7 runs took 23 ms of thread CPU time against
1,163 ms for the search.
Every click, hover, Select Behind list and marquee made new symbol boxes
and measured the art of each symbol its instances show again, and so did
the locked-artboard check each of them made. With 20,000 instances of
8,000 different symbols a click took 27 to 33 ms.
Hit tests now start from the art boxes, nesting depths and name index
that earlier hit tests measured on the same thread, for the last two
sets of symbols. The locked-artboard check of a hit test starts from the
same art boxes, and what it measures is kept with the rest. The boxes
are measured again when a symbol's name or art, the 9-slice guides or
the inputs of clipping-path effects change. The kept list holds a weak
pointer to each symbol's art, so it does not keep a closed document's
art alive, and art changed in place no longer matches. Art boxes of
symbols whose art holds 9-slice instances are measured on each call, as
before.
When the import refuses a file's PDF part and its editing data can't be
read either, the error named only the PDF part. It now adds "its editing
data couldn't be read either (<why>)", as layers already do
(crates/eps/src/import.rs). The reasons come from the editing data
reader (a size limit, damaged compression) and from the private data
streams (larger than 256 MB, another filter or decode parameters,
damaged zlib data). A file without editing data, and an open with pages
picked, layers off or the editing data off, keep the PDF part's error.
Part of #4366.
select.anchorsMany, which the Direct Selection and Group Selection
marquees, the Lasso and Smart Pen send, checked and changed membership
through Selection::contains, add and remove for each item, each a scan
of the selected objects. It now keeps the selected objects in a map
while the items run and writes them back once. The objects, their
order, the anchors, segments, key, key anchor and the other selection
fields come out as before, repeated ids and ids deselected and
selected again included.
Tests compare the result with the item-by-item loop for every mode,
four starting selections and fifteen item lists, and bound the passes
over the selected objects. An ignored benchmark times both loops on
2,000, 20,000 and 50,000 rectangles.
The harmony strip and the variation cells start a paint drag with the
primary button. The drag carries the grabbed color as the cell shows it
(snapped to the Limit to Library colors), and the Swatches panel's
existing drop makes a swatch of it with swatch.new: one undo step, the
color's own model, a drop on a color group joins that group, and a color
that is already a swatch is added again under another name. Starting a
drag does not apply the color or change the base or the selected cells,
and a plain click still applies the color as before.