Threaded DRC access will write to this string, re-allocating the memory
without any synchronization between threads using the string. Comment
adding this listed performance as a reason for using shared strings.
Measured performance does not seem noticeably different in either case,
even with high-error count boards. If there is a case where the
performance is limiting, we can replace these wxStrings with
std::wstring and utilize fmt
Fixes https://gitlab.com/kicad/code/kicad/issues/9888
(cherry picked from commit 5327b10064)
parentheses are rarely if ever used independently, so trim the unmatched
ones regardless of whether we find the starting parenthesis in the
description
Fixes https://gitlab.com/kicad/code/kicad/issues/11403
(cherry picked from commit be7e9d300e)
Better fix for 11750. Instead of figeting with layer values, we sort
our map based on the layer numbers in the PCAD file. F_Cu is always
layer 1 and B_Cu is always layer 2.
Fixes https://gitlab.com/kicad/code/kicad/issues/11750
(cherry picked from commit ac0f95683f)
- ensure the 'chamfer' radius and clearance is rounded before computing the actual hull vertices
- for small, non-45 degree line segments (present in many boards), generate a slightly larger but still axis-aligned 45 degree hull instead of following exactly the segment direction
(cherry picked from commit bcfe12ce47)
This is likely an old workaround to numerical precision issues
in the DRC system that no longer exist. Removing this is necessary
for "exact" hull generation to produce walkaround results that
look nice (paths falling exactly on the grid lines)
Fixes https://gitlab.com/kicad/code/kicad/-/issues/10710
When we generate hulls, by default we subtract the clearance epsilon
to prevent false collisions in the router that wouldn't be flagged
by DRC. However, we need to use the actual hull with no epsilon
when generating hulls for pathfinding in the walkaround system.
Without this change, it is possible for the walkaround to generate
a valid-seeming path that results in a DRC violation, for example
when dragging a drag against a board edge.
Fixes https://gitlab.com/kicad/code/kicad/-/issues/10536
Fixes https://gitlab.com/kicad/code/kicad/-/issues/11365
Fixes https://gitlab.com/kicad/code/kicad/-/issues/10710
According to the PCAD specification, all tags are case insensitive.
Most exporters honor the defaults but there is always an outlier or two
Fixes https://gitlab.com/kicad/code/kicad/issues/11652
(cherry picked from commit 35b79c2ec9)
"Highlight on PCB" and "Select" - "Items in Same Hierarchical Sheet"
actions were extended to select items on subsheets too.
"Highlight on PCB" now uses a proper full sheet path, allowing it to
work in subsheets.
Fixes https://gitlab.com/kicad/code/kicad/issues/11493
When loading footprints/zones in parallel, we cannot reference a static
vector that gets cleared by other threads.
It is unclear why this element was ever static to begin with (premature
optimization?) but it has been this way for a long time. We never
noticed until we threaded load cycles, which gave the hatch a chance to
override itself and cause crashes
Fixes https://gitlab.com/kicad/code/kicad/issues/9888
(cherry picked from commit 8218f9ab44)
Premature optimization appears to be the issue. When we threaded zone
handling and added multiple layers, we started overwriting the
pointbuffer variable, which was declared static for unknown reasons
Fixes https://gitlab.com/kicad/code/kicad/issues/9888
After changing track width, via size, grid or zoom from the toolbar, the
canvas has no longer the focus on Windows. So we force the focus after clicking
on these widgets.
From Master branch
Fixes#11103https://gitlab.com/kicad/code/kicad/issues/11103
1) Import leaders
2) Drop radius and angle dimensions
3) Import dimension text sizes
4) Correctly interpret which side dimension offset is on
5) Correclty handle offsets for non-cardinal dimensions
Fixes https://gitlab.com/kicad/code/kicad/issues/10763
The basic problem is that the DRC engine does length testing and skew
testing by collecting all the diff pair constituent parts and pairing them
itself. Since each part is collected on its own, we need to ignore the
'B' unit when evaluating any conditional expressions. However, doing this
in general means that when evaluating "OwnClearance()" we also ignore the
'B' unit and return the diff pair CLEARANCE_CONSTRAINT when we shouldn't.
This implements a more discerning test which know what the current requested
constraint is when evaluating expressions.
See also https://forum.kicad.info/t/solved-custom-differencing-rule-not-working-drc/34034/6
Fixes https://gitlab.com/kicad/code/kicad/issues/11314
(cherry picked from commit f7cdc7af75)
A triangular trapezoid will have the deltaX/deltaY of the same size as
the length of that axis. This creates a three-point polygon, which is
perfectly valid for our use
When building the connectivity database, we should not be using the
connectivity to check for shapes.
To make this deterministic, we introduce two flags (ALWAYS_FLASH and
NEVER_FLASH) that are used with connectivity building to determine
whether a pad is flashed for connectivity or not. ZONE <-> PAD/VIA
connectivity will be checked with ALWAYS_FLASHED and all other
connectivity will be checked with NEVER_FLASHED if they are marked for
potential annular ring removal. If they are not marked for removal,
they will be checked ALWAYS_FLASHED.
Fixes https://gitlab.com/kicad/code/kicad/issues/11114
(cherry picked from commit 442aae19d9)
Allow falling back to the alternate variable definitions if the built-in
text block variables are not set.
Also push the project into variable resolution when plotting
Fixes https://gitlab.com/kicad/code/kicad/issues/11168
Arcs can be altered by the process of changing from on-disk
representation to in-memory representation. Saving back to disk without
modifying the arc should not modify the calculated values.
This stores a copy of the on-disk representation that is only used to
save back to disk in the event that the arc is not modified during
editing.
Fixes https://gitlab.com/kicad/code/kicad/issues/10442
(cherry picked from commit cd7141fd10)
Avoid excess writes to the project directory to dodge issues with remote
file systems not fully writing data.
wxRename works across disk boundaries (in theory) and falls back to
wxCopy/wxRemove when it fails
Fixes https://gitlab.com/kicad/code/kicad/issues/10747
1) Drop tiny islands which might cause connection problems.
2) Cleanup some dead code in filler
3) Don't estabilish a connection to an island that is fully covered
by whatever its connecting to.
Fixes https://gitlab.com/kicad/code/kicad/issues/11036
The condition that the board must have elements on it before allowing
the rotate command is invalid as we allow rotating new elements before
placement
Fixes https://gitlab.com/kicad/code/kicad/issues/10581
If we are converting to tracks, the copper layer needs to be selected
before any other elements are processed, otherwise, we end up with the
default undefined layer
Fixes https://gitlab.com/kicad/code/kicad/issues/11025
The move even dereferences m_enditem for logging and perhaps other
actions. We need to ensure that the end item has not been invalidated
before taking this action
Fixes https://gitlab.com/kicad/code/kicad/issues/10732
Also makes sure the progress dialog is closed when we're done reading
symbols (it used to stay up for much of the symbol editor initialization).
Also makes sure that any cancel in the preLoad step is honoured in the
sync step. (The preload is done because it is multi-threaded and therefore
faster than the single-threaded sync.)
Also makes sure that individual threads pay attention to the cancellation,
not just the GUI thread.
Fixes https://gitlab.com/kicad/code/kicad/issues/8372
(cherry picked from commit 1f16092e29)
KiCad has stricter requirements for what consititutes an annotated
symbol. This checks for and corrects annotations that begin with '#',
do not begin with a non-numeric character or do not end with a number
(cherry picked from commit 78620347c8)
Fixes https://gitlab.com/kicad/code/kicad/issues/10760
Board BBox uses the visible size of the board, which includes edge cut
thickness. This is not appropriate for the Board characteristics.
Using the outline polygon gives accurate results and falls back to a
bounding box in the case where the outline is invalid
Fixes https://gitlab.com/kicad/code/kicad/issues/10972
*Fixes in the SVG import polygon postprocessing:
- don't drop subsequent polys when a non-filled primitive is imported 'in between'
- fix missing holes (also related to the interleaving of stroke and filled shapes, depending on the software that wrote the SVG file)
*GRAPHICS_IMPORTER_BUFFER: forgot to store the indices of the paths belonging to a compound shape
*pcbnew: GRAPHICS_IMPORTER can now recognize multi-path shapes (and postprocess polygons with holes into Kicad-compatible fractured polysets)
*SHAPE_POLY_SET: polysets can now be built from a bunch of arbitraily ordered oriented outlines.
Used by the SVG hole support
*SVG_IMPORT_PLUGIN: enable postprocessing of polygons with degeneracy and holes
*pcbnew: SVG importer support for even-odd fill rule
We should never be editing the user's outlines for them. This creates
broken zones that are hard to repair. In this case, there was also a
secondary zone with no outline that would crash the system
Fixes https://gitlab.com/kicad/code/kicad/issues/10904
When merging co-linear tracks, we should maintain the existing break if
it exists on pads, vias and other connection points
(cherry picked from commit 3a29d94c0a)
When merging segments that loop back on each other, we need to merge
them in such a way to to keep the resulting endpoints connected to the
original adjoining endpoints
Fixes https://gitlab.com/kicad/code/kicad/issues/10780
Now the netlist must be explicitly selected and read instead of beeing
read during dialog creation.
It also avoid a useless netlist read when closing the dialog or trying
to change the netlist filename.
From master branch
Also fixes a bug where zones meeting at a point would get merged
resulting in a self-intersecting zone.
Also fixes a bug where undo would not be handled correctly when zones
could not be merged.
Fixes https://gitlab.com/kicad/code/kicad/issues/10466
(cherry picked from commit d5a2059c21)
Handles checking output of the arc to ensure we don't end up generating
an invalid arc. Also keeps the limit of the arc angle to be (360,360)
excluding 0.
Fixes https://gitlab.com/kicad/code/kicad/issues/10070
If we wanted to put it into the parent's undo item, then the *parent*
would need to not start a new undo record. Not starting one in the
zone filler tries to add it to the *previous* change, not to the parent
change.
Fixes https://gitlab.com/kicad/code/kicad/issues/10091
(cherry picked from commit 4cc1ced2f3)
When setting previous values, we don't get the "OnFocus" event, so the
evaluation would not get triggered by simply repeating (using the
keyboard)
Fixes https://gitlab.com/kicad/code/kicad/issues/10752
gcc doesn't need it and computes during compile regardless, but experimenting in godbolt, both clang and MSVC actually do need it or else it's partially computed at runtime
(cherry picked from commit ca7840334c)
We now provide arc end points and interior angle, so we cannot check for
null radius in the same fashion. Instead, the starpoint==endpoint
indicates a zero radius
Fixes https://gitlab.com/kicad/code/kicad/issues/10714
WxString does not allocate space for wide strings needed during
conversion unless the string is explicitly wide. This can cause buffer
over/underflow
Fixes https://gitlab.com/kicad/code/kicad/issues/10605
jean-pierre charras
Seth Hillbrand
Jeff Young
Tomasz Wlostowski
Jon Evans
Alex
Marek Roszko
aris-kimi
Frank Zeeman
Steffen Mauch
Greg Davill
qu1ck
Mikolaj Wielgus
Graham Keeth
木 王
Roberto Fernandez Bautista