Direction Handles cleared the selection and then added each path on its
own, and each add searched the paths already selected, so building the
selection took time that grows with the square of the path count. The
command now selects the paths with one Selection::set call and adds the
anchor sets in one extend. The selection is the same: paths keep their
order, a repeated path keeps its first place and its last anchor set,
and the side fields are cleared as before.
crates/engine/examples/direction_handles_bench.rs times both ways on the
same paths in one process. With 20,000 four-anchor paths, building the
selection takes 1.9 ms of thread CPU time against 69 ms one path at a
time.
Cut Path at Selected Anchor Points cleared the selection and then added
each cut path in turn, and each add searched the paths already added, so
selecting the cut ends took time that grows with the square of the number
of cut paths. The command now selects them with one Selection::set call,
which skips repeats through a hash set and keeps the order in which each
path first appears, and then adds each cut end's anchor. The selection is
the same in every field; a new test compares it with the one-by-one build
on open, closed and multi-subpath cuts, long lists and an empty list.
The gain shows on large documents. With the 8 middle anchors of each of
5,000 open paths of 10 anchors selected, building the selection of 40,000
cut ends takes 9.7 ms instead of 232 ms of CPU time, and with 2,000 paths
3.2 ms instead of 39 ms (medians of 7 runs of the new
cut_path_select_bench example, both ways in one process).
A dashed stroke of a path with more than 1,024 path elements was outlined
whole in every frame after an edit. It is now outlined in and around the
view like a solid stroke. dash_kept runs the pattern along every segment
and keeps only the dashes and dots on the segments near the view, so they
sit where dashing the whole path puts them and the pixels are the same.
A pattern too fine to draw is still drawn solid, and that solid stroke is
cropped too.
At 1600 x 1200 the frame after one anchor moves takes 14 ms instead of
102 ms on a filled closed path of 50,000 anchors with 6 pt dashes and
4 pt gaps. anchor_bench now includes that document.
The renderer outlined the whole stroke of an edited path on every frame,
so dragging one anchor of a path with many anchors re-outlined every
segment. A stroke line of more than 1,024 elements without dashes or a
width profile is now outlined only for the segments within the stroke's
reach of the view, inflated by the view's own size on each side. The
cached outline keeps that region, and a frame outside it outlines the
stroke again. Closed subpaths that a wide stroke crosses over stay whole,
and a cut closed subpath is walked from a removed segment so no cut lands
where it starts and ends. The pixels are the same.
At 1600 x 1200, the frame after one anchor moves takes 28 to 30 ms
instead of 103 ms on an open path of 200,000 anchors, and 10 ms instead
of 30 ms on a closed path of 50,000 anchors (the new anchor_bench
example: cargo run --release -p vectorcraft-engine --example anchor_bench).
Writer::fresh_id probed prefix-1, prefix-2, ... from 1 on every call,
formatting and hashing each candidate, so writing N gradients, clip
paths or masks cost O(N^2). Ids are never released, so every number
below the last one handed out for a prefix is taken: remember where
the search for each prefix stopped and resume there. The ids written
are the same as before.
examples/still-life.vectorcraft (12k gradients, 6.3 MB SVG) exported
in about 8.6 s (median); it now takes about 0.15 s. Output is byte-identical
for every document in examples/.
Add crates/engine/examples/export_bench.rs to time document.export
on the example documents, and a test pinning the generated id
sequence when object names already take some of the numbers.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>