It was broken by commit c78171d01f.
During creation, the Drc must only display a error message, and do not create markers.
Markers are created only during a full board DRC test.
commit c78171d01f replaced this behavior by a incorrect one (create markers and do not show error messages)
The progress reporter should not be cleared in the footprint
implementation to prevent dereferencing a NULL.
When cancelling the load, we should only allow setting of cancellation
to true to prevent a condition where the progress reporter resets
the cancellation.
Fixes: lp:1774777
* https://bugs.launchpad.net/kicad/+bug/1774777
Also lock the DeleteAll call with mutext to protect threaded access.
The OpenGL contexts are removed when OPENGL_GAL class is destroyed.
Explicitly deleting all contexts prior to destroying the OPENGL_GAL
class causes an assert when the destructor tries to lock its context
prior to cleaning up OpenGL memory. In most cases, an unhandled assert
in a destructor-called function will simply exit the destructor.
Python thread cleanup will also attempt to close the context. This can
cause a race condition with multiple threads accessing/deleting the
canvas.
Fixes: lp:1774096
* https://bugs.launchpad.net/kicad/+bug/1774096
Colors are assigned to virtual layers while the ToLayer_ID returns a
physical layer enum. The GetLayerColor/SetLayerColor take LAYER_NUM
(int) values that can be larger than the physical layer count.
We were trying to handle aOnlyOneFile at two different levels
which was preventing the Silk layer from going through the
right routine (and preventing board outlines from getting
drawn if selected).
We were also overwriting the references color local variable
before using it, meaning the reference was always rendered in
the value's color.
Fixes: lp:1774171
* https://bugs.launchpad.net/kicad/+bug/1774171
Previously, the dialog was created only once, with the current board setup and the current units selection.
Therefore, after changes in board or units, the dialog was really broken.
Now, the dialog is recreated when the plot command is run.
A perfect solution would create holes in the coverage map
for the zone edges and handles, but that's a lot more
work.
Fixes: lp:1773204
* https://bugs.launchpad.net/kicad/+bug/1773204
As of v5 footprint library reorganization (one repo per library concept
is replaced with a common repo for all libraries), the Footprint Library
Wizard may handle only the old libraries. It has been replaced with
a file browser to select only local files.
Fixes: lp:1772209
* https://bugs.launchpad.net/kicad/+bug/1772209
When starting async processes, we need to have a way to stop the process
before releasing memory. Descoping FOOTPRINT_ASYNC_LOADER while the
threads were still running could cause crashes depending on the memory
structure. To avoid this we define clear procedures for exiting a
running async processes and call these when exiting.
Fixes: lp:1772909
* https://bugs.launchpad.net/kicad/+bug/1772909
The SELECTION is a std::set. Numerical dereference of item n is O(n) as
the iterator is not random access. Therefore, a for loop using
numerical dereference is O(n!) and quickly slows.
We avoid this by storing items to remove separately while iterating and
then removing the items after we complete.
Fixes: lp:1692081
* https://bugs.launchpad.net/kicad/+bug/1692081
If you select a track with many segments (e.g. a length-tuned track) and
then select all connected items again ('U' -> 'U'), we would iterate
over all items in the selection and mark connections for each of the n
segments n separate times.
We avoid this by using the marked flag to show when the segment has
already been visited by the routine. This means that if the segment has
been checked for connections because it was connected to the previous
item, it won't be checked for connections again. However, a selection
that interleaves items from multiple connection segments will still
(potentially) be multiply checked as the BUSY flag is cleared each time
there is one not-BUSY track in the selection.
On large boards with high pad-count modules, searching each pad for hits
becomes expensive. We eliminate many of the pad searches by first
checking the module's bounding box before iterating over pads to look
for hits.
It can be useful to have similar class messages grouped together so that
error messages in a larger report don't get lost among the
warning/info/action messages.
This patch allows sorting of messages for dialogs that benefit from
organization. Default behavior of reporter messages remains
unchanged by this patch.
Fixes: lp:1772090
* https://bugs.launchpad.net/kicad/+bug/1772090
Freerouting-specialized menus are removed as FreeRouting is not
maintained or packaged for most OSes. Users can still import/export
SPECCTRA files in KiCad allowing them to interface with a custom
FreeRouting installation.
The graphical editor window for pcbnew and modedit have slight
differences, so different classes. But they should appear the same to
the user. This merges the differences into a dialog that looks the same
to the user.
Fixes: lp:1772389
* https://bugs.launchpad.net/kicad/+bug/1772389
Circles are defined by center and a point on their edge. This requires
the user to do extra math to figure out the size of the circle. The
patch allows the user to edit and draw circles using radial coordinates.
1) hide the status popup when the dialog is brought up
2) forward events from the popup to the canvas so hotkeys work
3) make sure diff pair radius is always 100%
4) fix greyed-out Miter style label
5) make L a hotkey so it works before you start dragging
Fixes: lp:1545856
* https://bugs.launchpad.net/kicad/+bug/1545856
We only need to iterate over the anchors when there are items that are
marked invalid. We check once in the item list and only if there are
invalid items to remove do we trim the anchor lists.
However, the connected items might still need to be trimmed, so we leave
this final step outside of the conditional.
This is a speed commit for large boards. Tracks and pads cannot connect
to elements that are not on the same layer. Rather than checking for
this at the last step, this commit splits the anchor vectors by layer,
limiting the initial search space.
Previously, binary search was hand-coded. This moves the search to a
std::algorithm variant. Also searches bbox by limits rather than
directly iterating.
Maciej Suminski
jean-pierre charras
Seth Hillbrand
Jeff Young