It'd be nicer to know what happens, but we don't have a reproducible
case, so this "fix" will at least prevent that particular crash.
(cherry picked from commit 0091b9f897)
(They were only getting drawn on top of the selection shadow when
drawn by their parent symbol because LIB_FIELD was correctly
setting the view layers but SCH_FIELD was not.)
(cherry picked from commit 2300b0d2a3)
This commit partially reverts deb3cbf8. VIA and SOLID operator='s introduced in it do a good job but the cloning of holes in addHole() causes bogus collisions.
The diff-pair placer creates DIFF_PAIRs on the stack, which have
VIAs in them, which have HOLEs in them. The HOLEs get put in the
NODE's index, and we then crash later when doing a collision after
the stack variable has gone out of scope.
The stack variable is also copied while doing a search for "best",
and without the operator= definitions the "best" and the "original"
both think they own the HOLE.
Fixes https://gitlab.com/kicad/code/kicad/-/issues/14852
This was causing intermittent shove hiccups/freezes as RULE_RESOLVER::GetClearance() could return a bogus clearance value, taking
a cached value for an item that does not exist anymore. If the allocator/stack accidentally reclaimed such item's address - we were getting
very nasty and difficult to reproduce misbehaviours of the shove algorithm. Hopefully this patch fixes this. More info in the comments.
Note to self - I'm still not fully convinced I want PNS::ITEM pointers as the cache key, probably a unique counter would be better here.
Don't open-code knockout text shape generation in several different
places.
Make sure triangulated knockout text gets clearance added when
specified.
Collapse duplicated footprint text item plot routine (they're no
longer any different from plotting pcb text items).
(cherry picked from commit c71cf21e2f)
Also fixes a bug where we didn't triangulate at all when the char
count didn't warrant the likely overhead of spinning up a
thread_pool.
And fix another bug where EDA_TEXT::GetRenderCache() wasn't using
the given font.
Also reverts using the cache for drawing-sheet text. The text
items are created from scratch from the data items each time
they're drawn, so there's never an existing cache to make use of.
Instead, we now check that the item is in the view, using a very
approximate bounding box generator (because even generating a
real bounding box shows up large in profiles).
And, lastly, fixes a bug where EndPos was never considered in
DS_DATA_ITEM::IsInsidePage().
Fixes https://gitlab.com/kicad/code/kicad/-/issues/14822
(cherry picked from commit f35a88ce0b)
1) Don't fracture font glyphs when generating them; we're going
to fracture during triangulation anyway.
2) Don't check for self-intersection when deciding to fracture.
It costs nearly as much as the fracture does.
3) Cache drawing sheet text.
4) Use the current font when checking for cache validity.
5) Parallelize glyph triangulation.
6) Don't invalidate bounding box caches when offset by {0,0}
7) Use the glyph cache when generating text effective shape.
8) Short-circuit NormalizeJustification() if its center/center.
9) Don't triangulate for GuessSelectionCandidates()
10) Avoid sqrt whenever possible.
11) Pre-allocate bezier and SHAPE_LINE_CHAIN buffers.
Fixes https://gitlab.com/kicad/code/kicad/-/issues/14303
(cherry picked from commit 4ed267394a)
Aliases in bus resolution were being resolved without a path, making
them effectively global nets. This applies the corrected path to the
bus members and adds a QA to catch this error
using a non Kicad font
PCB_PLUGIN::formatRenderCache() was considering all shapes in cache are polygons,
but this is not true: some items can be stroke shapes.
From stable branch
Fixes#14804https://gitlab.com/kicad/code/kicad/-/issues/14804
SCH_CONNECTION objects are temporary and can become stale any time the
connectivity is updated. Keeping them around to reference later is a
bad idea. Even if the object pointer is still valid in an SCH_ITEM in
the undo/redo buffers, comparing the pointer against another pointer as
a test to see if they are the same connection is not valid. Saving the
connection name is safe and ensures the connection is the same even if
the pointers differ.
(cherry picked from commit 831a6d55fc)
Trying to squeeze out some kind of gain by informing the OS we will read a file sequentially.
In particular our FILE_LINE_READER just uses fgetc until the end
Jeff Young
Alex
Tomasz Włostowski
jean-pierre charras
Seth Hillbrand
Marek Roszko
David Holdeman
Mike Williams
Wayne Stambaugh