This is a boyscouting commit to standardize the threading of zone fills.
We do not need to join threads after their completion, instead we simply
allow them to clean up their memory without blocking the user. This
also sets the maximum number of threads that may be created to the
number of zones being filled. More than this will only leave un-used
threads being created and immediately killed.
We also include the connectivity search as a phase in the fill progress
reporter. This was the case before but did not utilize the correct
maxsize, leading to stalled progress bar.
(cherry picked from commit 969e85daa3)
This commit finishes the removal of OpenMP from the KiCad codebase.
Removed in this commit are the OpenMP calls in 3d-viewer and
qa/polygon_triangulation as well as all references in CMakeLists.txt
std::thread is used instead for multithreaded computation
(cherry picked from commit f8784f30a8)
When editing the pcb, Ctrl-Click will highlight a net. This allows
de-highlight with the ESC/cancel action from the selection tool. Inside
the highlight tool, ESC reverts to the selection tool while keeping the
net highlighted.
Fixes: lp:1793842
* https://bugs.launchpad.net/kicad/+bug/1793842
(cherry picked from commit 518b53f78c)
Clang++ seems to be in left-field here as it processes a stringstream
1.0X1.0 >> double as a hexadecimal floating point ('X' character) and
throws an error rather than converting 1.0 and stopping when it reaches
a non-numeric character. This causes errors when exporting step on Mac,
which uses clang++ by default.
Adding spaces to the string is more explicit and doesn't break gcc. It
will be sufficient until we get rid of kicad2step as an external
utility.
Fixes: lp:1778564
* https://bugs.launchpad.net/kicad/+bug/1778564
The lock file code would reset the lock file every other time the same
board file was opened in Pcbnew in the stand alone mode. This created
the perfect toggle switch causing the board to be reloaded every other
time the same board was selected.
Fixes lp:1777599
https://bugs.launchpad.net/kicad/+bug/1777599
OpenMP is not available for macos, so moving to a std::threads
implementation brings platforms into shared code.
This also reduces the OpenMP overhead when computing connectivity and
ratsnests.
Fixes: lp:1780130
* https://bugs.launchpad.net/kicad/+bug/1780130
(cherry picked from commit c0f067bf90)
The CMake scripts calculate the install path for python scripts by
distutils.sysconfig.get_python_lib(
plat_specific=0,
standard_lib=0,
prefix=''
)
which generates a string in line with
lib/python2.7/site-packages
This string ends up in the CMake variable PYTHON_DEST and is used by the
install step as destination directory for pcbnew.py.
There has been a hard-coded assumption on the content of that string in
PcbNew which is not compatible with FreeBSD 11.1.
This commit eliminates the hard-coded assumption in PcbNew and reuses
the install path as known by CMake.
Fixes: lp:1777921
* https://bugs.launchpad.net/kicad/+bug/1777921
Triangulation of zones can be time consuming for large boards. This
adds a std::thread implementation of parallel zone loading to speed this
process on multi-core machines.
(cherry picked from commit a04ef37d67)
When running GAL, the connectivity is built when the board loads, so the
extra call to rebuild connectivity is not needed.
(cherry picked from commit 13b96799ea)
This prevents non-copper layers from being included in the connectivity
search. It also limits the layer search in the connectivity RTree to
just the copper layer range.
(cherry picked from commit f87cb64d65)
Repeating find with the same search string would result in a not found
error for absolute search strings. Don't clear the found item if the
search string did not change from the previous search..
Fix another minor bug where the do not warp mouse pointer option was
always checked when the find dialog was called from one of the gal
canvases.
Fixes lp:1793300
https://bugs.launchpad.net/kicad/+bug/1793300
Zones can be explicitly merged even if they do not have the same
parameters. Automatic merging should only occur when all of the zone
parameters are exactly identical to allow for area-specific zone
parameters.
Fixes: lp:1695761
* https://bugs.launchpad.net/kicad/+bug/1695761
Keepout zones can have different layer sets and different restrictions
that should not be combined unless they are identical.
Fixes: lp:1789062
* https://bugs.launchpad.net/kicad/+bug/1789062
We had been deselecting items after calling edits to be safe in case the
item was deleted/exchanged by the edit. The item pointer itself remains
valid even when deleting as it is assigned to the undo stack. But it
should not remain visible or selected on the schematic if it is removed.
This tests for removed items by checking whether it (in the case of
first-level BOARD_ITEMS) or its parent (in the case of footprint item
components) remain in the view list after editing. If they are still in
the view list, then we re-select them.
Fixes: lp:1765774
* https://bugs.launchpad.net/kicad/+bug/1765774
Fixes: lp:1775946
* https://bugs.launchpad.net/kicad/+bug/1775946
Snapping while dragging depends on the current net being assigned during
dragging. This initializes the placer to hold the current net while
performing inline dragging. The placer is reset along with the dragger
in StopRouting(). This fixes the difference in snap between drag and
create new route.
Allow corner adding to polygons in both module editor and board editor.
The board editor was already allowed but only through special types.
The module editor is now sychronized to the same process and both
recognize standard context menus
Fixes: lp:1782966
* https://bugs.launchpad.net/kicad/+bug/1782966
When constraining moves, the cursor snaps to grid but does not
represent the position of the final edited point. The point is rotated
to the nearest snap angle, while the vector length remains unchanged
resulting in a line longer than intended and off-grid.
This adds a grid-snap after the constraint is applied.
Fixes: lp:1792655
* https://bugs.launchpad.net/kicad/+bug/1792655
Normally working at a zoom level when a footprint is not fully displayed
indicates that the user modifies other components. Selection Tool used
to discard footprints that had area larger than the viewport area, but
it may make sense to select them if there are no other candidates.
Fixes: lp:1789578
* https://bugs.launchpad.net/kicad/+bug/1789578
(cherry-picked from commit 5993b330)
Unplated holes in Ealge can either be free floating or part of a larger
footprint. We handle both in the same packageHole routine by either
offsetting the hole in a centered footprint or offsetting the footprint
in a centered hole.
Fixes: lp:1791287
* https://bugs.launchpad.net/kicad/+bug/1791287
Invalid graphical items that cannot be created in pcbnew may still be
created by external tools. This synchronizes the pcbnew view display
with the gerber output and 3d viewer line widths for these types of
lines. Only filled polygons are permitted 0-width lines.
Fixes: lp:1790534
* https://bugs.launchpad.net/kicad/+bug/1790534
This allows full copper connections even when THT changes a trace layer.
It avoides selecting the actual pad to prevent actions on the copper
connections from affecting the footprints as well.
Fixes: lp:1789807
* https://bugs.launchpad.net/kicad/+bug/1789807
Items removed from the connectivity list will not have valid parents, so
we cannot update the BBox, thus we will only use the cached value.
Fixes: lp:1785605
* https://bugs.launchpad.net/kicad/+bug/1785605
Setting up the footprint viewer will attempt to load the current
footprint. If the library does not exist, the loader will throw an
IO_ERROR that needs to be caught.
Fixes: lp:1789626
* https://bugs.launchpad.net/kicad/+bug/1789626
Missing types were handled by the superclass (in the case of zones)
PCB_CONNECTED_ITEM, which incorrectly returned "true" when testing lines
against zones.
Fixes: lp:1789195
* https://bugs.launchpad.net/kicad/+bug/1789195
Before closing the window in the footprint editor, we clear the existing
pcb, undo/redo stack and associated memory to prevent callbacks
double-freeing on destruction.
Fixes: lp:1784504
* https://bugs.launchpad.net/kicad/+bug/1784504
The default board soldermask clearance should not be used in the
footprint viewer. Setting the board's solder mask margin to 0 will
leave the footprint displayed with only the soldermask defined by
the pads or the footprint.
Thanks to Fabian Inostroza for pointing out the missing line.
Fixes: lp:1779979
* https://bugs.launchpad.net/kicad/+bug/1779979
Perspective is overzealous: it stores everything, so we need to be sure these panes are not hidden,
regardless what the perspective has stored.
Fixes: lp:1789254
https://bugs.launchpad.net/kicad/+bug/1789254
Eagle files that are edited may be missing required data. While we do
not need to handle these files, we do need to prevent segfaults in KiCad
when reading them.
When changing elements in the module editor, each element shares the
same parent. Undo commits store a copy of the parent as it existed
before the change. For footprints with many elements, this creates
large, slow undo commits as a copy of the full footprint is stored for
each element being edited.
This keeps a single copy of the footprint in the undo stack per edit.
Fixes: lp:1780526
* https://bugs.launchpad.net/kicad/+bug/1780526
UI update event handler updates toolbar button state for the scripting
console, but for the associated menu item it displays an assert saying that
the menu item that cannot be checked. To avoid that, the UI update event
handler returns early if event does not come from the toolbar.
(cherry picked from commit 45c1742a)
In Cairo, we don't have z-ordering implemented, so we need to draw the
clearance outline first otherwise it will draw a large, grey overlay
that hides the actual track while routing.
Fixes: lp:1779228
* https://bugs.launchpad.net/kicad/+bug/1779228
After refilling zones, we need to unconditionally remove the filled
polys from the map because their fills map have changed. Left-over
filled polygons will not map to current polygons, creating "ghost"
islands in the connectivity. Removing these can overflow the polygon
count.
Fixes: lp:1778454
* https://bugs.launchpad.net/kicad/+bug/1778454
This adds a small KiCad logo in the top left of the Doxygen pages.
54px used as 55px is maximum recommended height for Doxygen logos.
For the main docs and the dev docs, this is easily done in the Doxyfile,
for the Python docs, as that is run from the binary dir, the source dir
is passed as an Environment variable.
The itemlist and zonelist in connectivity can both contain many items,
so both use the same OpenMP routine. However, we can only clear the
dirty flag when we conduct a full search, including zones. Otherwise we
missing connections to zones when propogating changed items that are
then marked as not dirty.
Fixes: lp:1777993
* https://bugs.launchpad.net/kicad/+bug/1777993
During the symbol library table implementation, the legal character
tests for the footprint library table were changed and the forward
slash '/' character became illegal. This change broke editing some
users footprint library tables that already had '/' in library table
nicknames. This change split the library nickname and library item
name illegal character tests.
The local ratsnest needs to be able to hide as well as show visibility.
Clicking a module will toggle the visibility while clicking off all
modules clears the local ratsnest display.
Fixes: lp:1531332
* https://bugs.launchpad.net/kicad/+bug/1531332
Options that do not have code implementing them are now hidden from the
user so that they are not searching for ways to enable the greyed out
options. Similarly, when selecting a routing option that disables
free-mode, we show the effect of disabling the mode in the dialog.
GAL is strict regarding which layers are rendered, therefore to show
ratsnest lines when an item is dragged - the ratsnest layer must be kept
enabled. Ratsnest line visibility is enabled per item and this is the
way to show/hide the lines.
Fixes: lp:1666473
* https://bugs.launchpad.net/kicad/+bug/1666473
Minimum differential pair clearance should be limited to the netclass
differential pair spacing. This also cleans up some unused code that
was left over from previous work on differential pair clearances.
Fixes: lp:1673632
* https://bugs.launchpad.net/kicad/+bug/1673632
Seth Hillbrand
Wayne Stambaugh
Martin Aberg
Jeff Young
jean-pierre charras
Maciej Suminski
Ronnie Gaensli
Carsten Schoenert
Baranovskiy Konstantin
Tomasz Wlostowski
John Beard