It was confusing that the primary frames registered their tools
differently than the other frames. In addition, since the other
frames also added their own tools, foo_actions::RegisterAllTools()
didn't really register all tool but rather those used by the
principal frame (PCB_EDIT_FRAME, SCH_EDIT_FRAME, etc.)
1 - For debug purposes, the code to fill non copper zones was disabled and the copper zone code was uses instead.
It creates issues with pad clearance areas, because a non copper zone has no clearance areas.
2 - The non copper zone dialog had no option to set the outline smoothing option.
Unfortunately, this parameter was inherited from the last edited copper zone.
And therefore the non copper zone can have a non wanted outline smoothing option.
This option is now added to the dialog.
Fixes: lp:1826952
https://bugs.launchpad.net/kicad/+bug/1826952
When Pcbnew's add footprint browser is run up, it constructs
the recently used list without checking for nulls.
Although the DoAddLibrary call does internally check for
null, it's easier to filter these out before placing into
the vector in the first place.
The same logic in the symbol tree is already handled in the
same way.
In this case, the history has a cleared link that created the crash when the dialog tried to select this deleted FootprintEnumerate
Fixes: lp:1826794
https://bugs.launchpad.net/kicad/+bug/1826794
In the failure case, the "makeType" is uninitialised. In a non-debug
build, wxFAIL will compile out and the uninitialised behaviour will
happen silently. In a debug build, it'll assert, and then do the
uninitialised behaviour anyway.
Continue explicity to avoid this.
Single pad nets deletion in dry run was incorrect when a new footprint was added, because in dry run it is not actually added,
and the single pad nets deletion does not see these new pads, and can (incorrectly) delete some nets.
This was always possible from modal browsers and from the footprint
editor, but it's now also possible from non-modal browsers and the
symbol editor.
Fixes: lp:920380
* https://bugs.launchpad.net/kicad/+bug/920380
Give the user the option of cancelling a file open if there are
segment zones; otherwise they're converted to polygon fills.
Fixes: lp:1823087
* https://bugs.launchpad.net/kicad/+bug/1823087
Move repeated GAL config reading routines into GAL_DISPLAY_OPTIONS.
THe app-level config is in here already, do the same for the
common config.
This means that the configs are loaded consistently, which fixes
the symbol-chooser preview window, which previously didn't use the
same config routine as other GAL canvases.
Future work could move these functions to free functions that
act on the public interface of GAL_DISPLAY_OPTIONS to avoid
GAL_DISPLAY_OPTIONS having to know about wxConfig and wxWindow.
Fixes: lp:1824524
* https://bugs.launchpad.net/kicad/+bug/1824524
Previously, the GAL_DISPLAY_OPTIONS object in FOOTPRINT_PREVIEW_PANEL::New
was passed by reference in the ctor, down to EDA_DRAW_PANEL_GAL, which stored
it as a reference. The object in New() then goes out of scope, so the
referencing panel outlives the options.
Fix this by making a copy in a std::unique_ptr of the options, and giving
ownership to the panel.
There is another issue here: when the Pcbnew options are copies, the
OBSERVABLE subscriber list is copied too. This means if the panel called
NotifyChanged() on the options, Pcbnew would get updates, even though a copy
of the options changed. However, the panel doesn't change the options or
notify, so it's OK for now.
The 3D viewer also uses OpenGL, so it also needs to
adjust based on DPI scale factors.
This patch moves the 3D viewer common config code solely
within the 3D viewer, and adds handling of the DPI scale. This
also avoids duplicated logic in PCB_BASE_FRAME::CommonSettingsChanged.
For now, as EDA_3D_VIEWER is not included in the Kiway settings
dispatch, still manually invoke the settings update from
PCB_BASE_FRAME.
The PANEL_3D_PREV gets a similar function to read the config, but
it doesn't have a listener for the update, so it's set just once
at dialog init, just like the existing pan settings.
Fixes: lp:1823681
* https://bugs.launchpad.net/kicad/+bug/1823681
This allows us to make the various netlist and pcb update routines
more atomic and less reliant on carefully sequenced asynchronous
messages.
This is also a prelude to adding support for footprint testing
without a netlist.
This behavior existed in 5.0.x but did not properly exit the edit tool
in 5.1. This removes stops the edit tool when we begin dragging a track
using the interactive router.
Fixes: lp:1820312
* https://bugs.launchpad.net/kicad/+bug/1820312
For legacy boards, the setting is picked up from the board outline
thickness. If the board outline has mixed thicknesses, then the
max is used and a warning is displayed.
Fixes: lp:1797787
* https://bugs.launchpad.net/kicad/+bug/1797787
There is a circular dependency between the GRAPHICS_IMPORTER object and
the GRAPHICS_IMPORT_PLUGIN object which makes sharing the settings in
the GRAPHICS_IMPORTER unwieldy. This fix is a ugly hack that resolves
the issue. The underlying issue will require some major refactoring.
Fixes lp:1821234
https://bugs.launchpad.net/kicad/+bug/1821234
Improve padding, spacing and alignment of dialog layout.
Fix title capitalization.
Make file dialog wildcard string translatable.
Remove fixed initial dialog size to allow sizers to work their magic.
Replace OK button event handler with TransferDataFromWindow.
Remember last line width units between dialog instances and sessions.
Save configuration file settings in a separate group and remove GfxImport
prefix from the entry names.
Add fencing to dialog header file to prevent nested includes.
Fixes lp:1822568
https://bugs.launchpad.net/kicad/+bug/1822568
Reverting changes causes the item to become invalid. This can cause
issues when we have dangling pointers for the VIEW routines left over.
Until we implement a method of passing the reverted element back up to
the calling routine, we need to remove the selection for each Revert()
call.
Fixes: lp:1821909
* https://bugs.launchpad.net/kicad/+bug/1821909
When removing a point from the polygon, the vertex is removed before it
is tested for validity. If it fails validation, the edit is reverted
and the point count doesn't change, so the view update only modifies the
existing points. But the revert may also have changed the shared
pointer to the point array, invalidating the view's copy.
Fixes: lp:1821909
* https://bugs.launchpad.net/kicad/+bug/1821909
This inserts a translation map in the layer names so that the old stable
files with Italian/French layer names get updated to the standard
English layer names (instead of not opening).
Fixes: lp:1370575
* https://bugs.launchpad.net/kicad/+bug/1370575
Save/restore default netclass.
Read layers back in with correct syntax ('.' between keyword and layer name).
Write layers to project file.
Read/write layer enablement bits.
Don't set solder paste margin min to 0 (most of them are negative).
Fixes: lp:1811990
* https://bugs.launchpad.net/kicad/+bug/1811990
the filled areas can use a hatch pattern (crossing lines) using square holes.
The zone filler removes too small holes (truncated holes) from hatch pattern.
It avoid to create small holes when a hole pattern is truncated by the filled area base shape.
Tab through the grid vertically, then horizontally, as the fields
are laid out with "related" fields (like the co-ords) vertically.
Also neaten headers.
Fixes: lp:1821887
* https://bugs.launchpad.net/kicad/+bug/1821887
Commit 6006703798 fixed a crash but broke the filled zones changes detection.
Filled zones were always seen as not up to date due to the fact the filled areas were cleared too early.
The up to date detection is also optimized: the old filled polygons are no longer stored.
Instead of, the MD5_HASH is calculated and stored before clearing the filled polygons.
Autosave was not working correctly, partly due to different
implementations in eeschema and pcbnew and partly due to a mistaken
refactor at some point during v5 development. This unifies the expected
autosave prefix to _autosave- for both pcbnew and eeschema. It also
unifies the expected suffix for the backup files to -bak.
Fixes: lp:1820433
* https://bugs.launchpad.net/kicad/+bug/1820433
A zone refill display progress bar is shown when deselecting the zone after geometry changes.
However, on Linux, there are some artifact between the progress bar refresh and the board view refresh:
the filled areas are incorrectly shown: the insulated islands remain displayed,
although they are removed from the actual filled areas list
So it is disabled until a fix is found.
When dragging, you will seldom want to snap to the original item. This
commit prevents the "size of track" snap to the original item and
replaces it with an auxilary grid origin for the snap. This means
effectively that the snap will happen to grid for the original item
unless the cursor is closer to the original item line than it is to the
grid crossing.
Fixes: lp:1820248
* https://bugs.launchpad.net/kicad/+bug/1820248
Previously, in many cases the selected zone was refilled on exit selection, even if the zone was not modified.
This is annoying, because the zone fill can be really time consuming, and cannot be called without a good reason.
Now the refilling is made only if a zone parameter is actually modified.
The bounding box calculation needs to be fast and is threaded in many
places in pcbnew. We cannot invalidate items (the function is const but
this wasn't caught at compile-time for some reason) when getting the
bbox. This shouldn't be an issue as the m_BezierPoints is rebuilt each
time the curve is updated, so rebuilding during the bbox seems extra.
Fixes: lp:1819984
* https://bugs.launchpad.net/kicad/+bug/1819984
Removing existing fills will prevent re-drawing invalid polygons before
they are ready. This seems to happen on Macs more easily due to redraw
timing.
Fixes: lp:1819553
* https://bugs.launchpad.net/kicad/+bug/1819553
Allows 0 to 4 chamfered corners, not only one.
A custom shape allow this kind of shape. However because it is a primitive,
it is easier to edit and it support thermal reliefs.
This was due to the stub removal tests if a stub end is outside a solid area to remove it.
There is a corner case: is the stub end is exactly on a solid area outline extra segment
created by Fracture() used in calculations.
The stub removal tests are now made before fracturing.
Fixes: lp:1819317
https://bugs.launchpad.net/kicad/+bug/1819317
The module refdes always increment by one. This is problematic
for two reasons:
* It doesn't work when more than one FP is selected (say you have
C1 and C2, you get C1, C2, C2, C3)
* It messes with simple panellisations
The 5.0.x behaviour is to not renumber in Pcbnew, so revert to
that behaviour.
Better refdes incrementation can come post 5.1, as this will need
new strings and UI.
Fixes: lp:1819106
* https://bugs.launchpad.net/kicad/+bug/1819106
Disable the text update event for the orientation cutom value field before
updating it when the radio buttons are selected. This allows to use the
conventional EVT_TEXT handler, rather than KEYDOWN, which has two effects:
* Tabbing though the dialog no longer sets custom orientation
* Click-driven clipboard actions trigger the update correctly.
Introduce a helper function in validators.h to assist in updating
validated fields without triggering further events.
Fixes: lp:1819006
* https://bugs.launchpad.net/kicad/+bug/1819006
When checking for closed outlines, we generate a board area using the
pcb structure. We now also check for overlapping outlines due to
lp:1818163 However, the DRC maintained its previous outline before
generating the new one, causing an overlap error.
This regenerates the board outline from zero for each DRC run.
Fixes: lp:1819012
* https://bugs.launchpad.net/kicad/+bug/1819012
pcbnew refreshes the viewItem for the worksheet multiple times. This
requires setting the filename to get parameters displayed on the
worksheet correctly.
Fixes: lp:1819000
* https://bugs.launchpad.net/kicad/+bug/1819000
Depending on thermal reliefs size, the polygons could be self intersecting,
and these self intersecting polygons are unexpected in zone calculations.
Other fix: the width of thermal stubs is clamped to the size of pads.
Previously, a bug in clamping calculations prevents constraining thermal stubs width.
Fixes: lp:1818752
https://bugs.launchpad.net/kicad/+bug/1818752
Set the focus to some control in the copper zone, non-copper zone, and
keep out zone dialogs required for the default button action to work
properly. This only seem to affect windows dialogs.
Fixes lp:1818397
https://bugs.launchpad.net/kicad/+bug/1818397
wxWidgets has several bugs that result in cell editors being closed
right after they're opened. There are two wxWidgets hacks to
partially address this: the SetInSetFocus() hack, and a slow-click
hack. We used to try and work-around these bugs ourselves for
single-click access, but this changelist moves those over to
wxWidget's slow-click hack.
Fixes: lp:1817965
* https://bugs.launchpad.net/kicad/+bug/1817965
Added checks to StrCmpNum() function in common/sting.cpp to prevent
iterating past the end of the sting. Also removed length of string
parameter since it did not seem to be used anywhere.
Fixes lp:1818157
https://bugs.launchpad.net/kicad/+bug/1818157
* fp editor: ensure axis are drawn after switching the backend
* DIALOG_PNS_SETTINGS: ensure violate DRC option is correctly enabled/disable when opening the dialog.
Violating the DRC cannot happen in shove or walk around mode as we use
the DRC to calculate these actions. The tooltip explains this as well
but we need to explicitly disable the checkbox to prevent users from
thinking that they can complete DRC violating tracks in other modes.
Introduce a shortcut method to set tab ordering and use it in
the dialogs that override the tab traversal orders:
* Eeschema sheet properties
* Pcbnew Move exact
* Pcbnew Move relative
Also set some initial focus fields in the same dialogs.
Tidy a few includes.
Fixes: lp:1816009
* https://bugs.launchpad.net/kicad/+bug/1816009
Includes Edit Text and Graphics Properties and Edit Track and Via
Properties. It's also no longer board-specific as it turns out the
board timestamp is always 0 anyway.
When moving user user units out of the global, we missed that Python
calls were using it. This re-adds a function to retrieve user units
allowing the python scripts to show users dialogs in their preferred
units.
The default hot key definition for toggle high contrast mode was 'H'.
On windows, the menu accelerator handler code was grabbing the character
so it would not appear in the filter text control. This is expected
because the windows wxWidgets code hooks this at the application level
so that once a menu accelerator key is found, all further event handling
for that key stops. Changing the hot key definition to 'Ctrl+H' resolves
the issue.
Fixes lp:1803556
https://bugs.launchpad.net/kicad/+bug/1803556
The best size was calculated from a default size.
This default size (calculated by wxWidgets) was fully broken with some texts in a few languages.
So now this best size does not use this default size.
Fixes: lp:1816315
https://bugs.launchpad.net/kicad/+bug/1816315
gal: pixel alignment for Cairo.
gal: pixel alignment for OpenGL
Cleanup of pixel alignment patches
gal/opengl: more pixel alignment patches:
- works on Retina displays now
- pixel alignment for circles
- correct 1-pixel strokes
- fixed unfinished 1-pixel line ends
GAL: Restore antialiasing options in Cairo
GAL: Fix arc drawing
Removes DrawArcSegment, which drew arcs whose edge was optionally
stroked (a feature that we did not use). Fixes Cairo arc drawing issue
where arcs were not visible in pcbnew.
gal: further cleanup
gal: removed unused shader parameter
Auto-selecting of the first grid element is performed in the footprint
editor to facility rapid editing. Rapid editing of the 3d-model fields
is not as useful and requires multiple esc-keys to cancel. This limits
the auto-edit to only the reference tab.
Fixes: lp:1815456
* https://bugs.launchpad.net/kicad/+bug/1815456
Highlight connections is meant to show the copper connections on a
board. We remove the non-copper pads from our selection when attempting
to source the net for highlighting.
Fixes: lp:1815898
* https://bugs.launchpad.net/kicad/+bug/1815898
Items that are located at the edge of or outside of integer limits will
not be shown on screen as they don't fit in the VIEW RTree. This
prevents them from being edited and can corrupt data output. Instead,
we place them at the largest value that is representable on the RTree
(INT_MAX/SQRT(2)). This allows the user to correctly utilize DRC as
well as interact with and remove the offending items.
Fixes: lp:1815642
* https://bugs.launchpad.net/kicad/+bug/1815642
The parent window was always the board edit frame,
but this is incorrect when this dialog is called from the Plot dialog.
Fixes: lp:1815530
https://bugs.launchpad.net/kicad/+bug/1815530
When adding colors to length tuner, we depend on the window background
color, which can be set by the user, resulting in an unreadable text.
This uses the HSL conversion to detect which shade it should use for the
coloring.
Also adds test cases for round-trip HSL and HSV conversion
Fixes: lp:1814530
* https://bugs.launchpad.net/kicad/+bug/1814530
Update 3D model panel on initial dialog load to reflect the current
visibility state of the 3D model.
Fix a few minor footprint properties dialog spacing issues.
Fixes lp:1815184
https://bugs.launchpad.net/kicad/+bug/1815184
Fix wxGrid column minimal widths on Windows in dialogs (just fixing the size is not enough).
WX_GRID::GetVisibleWidth(): make width bigger for labels, because they usually use a bold font.
(However, the fix is just a workaround: a better code is welcome)
Using the normal command, oval holes are "drilled" by a route command.
Some board houses do not accept this normal command.
In this case, the drill dialog allows using an other Excellon drill command (a slotted hole).
Note also some board houses do not understand this alternate command!
GTK does not handle enlarged text layouts automatically, so we need to
get the screen size before choosing the correct grid width for the
dialog
Fixes: lp:1788495
* https://bugs.launchpad.net/kicad/+bug/1788495
When doing a global edit of the tracks and vias, we need to allow for
the possibility that the track size doesn't change and that this is
explicitly allowed without DRC error.
The using the "first available" numbering option, force an
artifical linear numbering scheme starting at '1'.
Start the pad name provider at the 0'th pad index.
Also adds a few related tests and adjusts tests affected by
this change to reflect that the offset still applies.
This can be expanded in future to be more general by skipping
pad (which would allow a custom start and numbering scheme
while still avoiding duplication), but for now,
this does what the UI says and avoids string changes in 5.1rc.
Fixes: lp:1814918
* https://bugs.launchpad.net/kicad/+bug/1814918
Fixes: lp:1814917
* https://bugs.launchpad.net/kicad/+bug/1814917
* Make LIB_TABLE_BASE::GetCount() return unsigned. This is more
consistent with the behaviour of STL containers (especially the
boost::ptr_vector this is really accessing). Sadly
wxGridTableBase() forces an int, so a cast is still required
at the WX interface.
* Make LIB_TABLE_BASE::At() return a reference. First, this is more
consistent with normal STL indexing operator[]'s, and secondly, it
allows an idiomatic const index method (so you can access const
LIB_TABLE_ROWs from a const LIB_TABLE_BASE).
The motivation is to allow use of this class and LIB_TABLE_ROW
in a test program, where the LIB_TABLE_BASE is const.
Libpolygon can provide its own includes via target_include_dirs PUBLIC.
This means any linking targets do not need to specifiy them manually.
As common requires polygon, the polygon dep is also now no longer
required downstream of libcommon, as it's transisitvely implied
by libcommon's target_link_libraries.
This resolves a circular dependency previously detected and also
simplifies CMakeLists.
The bitmap definitions (BITMAP_DEF and so on) do not
have any dependencies on other libs, including WX. This
means the bitmaps library can be isolated from the other
dependencies.
Common now depends on bitmaps, and libraries that depend
on common can pick it up from the common target_link_libraries,
as it is PUBLIC. This means a lot of targets no longer
need manual bitmap linkage.
This avoids a circular dependency that was previously reported
by static analysis.
Avoiding pulling in WX and other headers into the include
tree of each bitmap .cpp is a huge speed up (around 10x) in
compilation, and the generated static library is also 10x
smaller (20MB vs 200MB)
Add common as a link library to pnsrouter,connectivity.
THese library do still use common code (including bitmaps,
via base_screen.h) and this allows them to pick up the libcommon
includes correctly.
In countries using a comma as floating point separator, the --min-distance="0.0100 mm" was passed as --min-distance="0,0100 mm"
This is incorrect and broke the command line.
Fixes: lp:1814559
https://bugs.launchpad.net/kicad/+bug/1814559
Circles and Squares are handled correctly. Octagons are converted to
circles. Pad 1 is identified by the "1" string in the Eagle name.
Fixes: lp:1784126
* https://bugs.launchpad.net/kicad/+bug/1784126
Fixes and issue where forces were calculated using the current track
width even in differential placer mode where the tracks are specified by
the differential pair width.
Fixes: lp:1814480
* https://bugs.launchpad.net/kicad/+bug/1814480
When creating new lib table rows, the pointer becomes managed by the new
table through the boost::ptr_vector. The row is non-copyable, however,
so we lose the reference as soon as it goes out of scope. Creating a
clone provides a new pointer that is moved into the table on creation.
Fixes: lp:1792456
* https://bugs.launchpad.net/kicad/+bug/1792456
Choosing the drag origin should not be dependent on the snap settings in
pcbnew. Snap settings are sensible when you are designating a target
only. Additionally, when dragging a module, we do not want the
footprint's pads to be used as targets for snapping.
Fixes: lp:1814402
* https://bugs.launchpad.net/kicad/+bug/1814402
When the user cancels the footprint load, we should assume they are
canceling the placement of a new footprint. This also adds sanity check
when populating the list
Fixes: lp:1814181
* https://bugs.launchpad.net/kicad/+bug/1814181
It was a regression from 5.02 version.
The block of new footprints to place has now its reference point (the point that is on the graphic cursor) fixed.
This is the first footprint loaded from the list.
This breaks the following functions out to a general-purposed refdes utils
header:
* MODULE::GetReferencePrefix()
* kicad_string.h RefDesStringCompare()
This acheives:
* Slimming of MODULE interface
* Placement of refdes code in common rather than pcbnew
** Testing of this code in qa_common
* Tighter and smaller includes for code that only needed refdes functions
Note: there are failing tests commited (as expected failures). These
are the cause of lp:1813669 and will be fixed as a follow-up commit.
This is not logic specific to MODULE. Breaking it out to
kicad_string.h acheives:
* Slimming of the MODULE interface
* Enables reuse of the function
* Enables testing of the function
Also add a test under qa_common for this function.
The CONFIG_SAVE_RESTORE_WINDOW class does not need to be tied
to the array dialog, put it in common/widgets.
Also do a refactor and tidy-up of the the class, use a union for (slightly)
better type-safety and syntax (a variant would be better but that's C++17).
Also enable integral field save/restore from text boxes.
Fully initialise the CREATE_ARRAY_DIALOG_ENTRIES object,
and use these values for the dialog on first open. Currently,
the initial values aren't all initialised, and none are used, as the
valid flag is not set. This means there are poor defaults in some
fields.
Also more declarations and defintions of CREATE_ARRAY_DIALOG_ENTRIES
to the .cpp, as these do not need to be exposed in the header,
even privately.
The original item is part of the array, so give it a number
and put it in the right place (all current array tools have
a null transform for the first item, but it's not part of the
contract. For example, a circular array with an angle offset
might want to move the first point).
The numbering axis priority (h then v or v then h), and alternate
reversing are not actually numbering parameters, they are part of
the geometry of the array (the items are physically in different
orders).
So enable these options, even when numbering is disabled.
Also fix broken logic about when numbering scheme is enabled
(it still applies even if the numbering offset is fixed.)
The current module cannot be queried for existing pads as we
add them (because we never commit, until we finish the whole array).
Instead, pre-gather the names and check as we add, skipping any existing
names.
Note: this will not prevent arrays becoming "mis-ordered", but there
is not a lot we can do to prevent all possible errors.
Fixes: 1808706
* https://bugs.launchpad.net/kicad/+bug/1808706
The same principle could be used to skip existing ref-des'es on PCBs.
The geometry and numbering logic is separate to the dialog, and
can be separated for clearer logic and better testability.
Moreover, refactor it to avoid any dependency on pcbnew
classes, so it can be move to common for potential re-use in
eeschema and friends in future.
Also convert all wxPoint logic in these classes to VECTOR2I and
fix some function visibilities.
Add some unit tests of the ARRAY_OPTIONS geometry and numbering.
Snap/Grid modifier follow preferences from 05483a06c6. Brings Shift/Alt
modifiers into alignment with drawing tools. Shift modifier still
conflicts when choosing end item with the complete track click-modifier.
This will be addressed in v6 with a larger preference setting.
Snapping to item options were partially disconnected when unifying
framework. This re-attaches the options and adds the additional option
for graphical items.
Fixes: lp:1801377
* https://bugs.launchpad.net/kicad/+bug/1801377
kicad2step has a tolerance level that sets whether two points are at the
same location or not. This allows KiCad to access these levels to some
degree, permitting a looser interpretation of a closed board outline.
The selection for interactive drag uses cursor position off grid to
choose the element for dragging. We pass the same value into the start
dragging routine to choose the correct end of the element.
Fixes: lp:1813665
* https://bugs.launchpad.net/kicad/+bug/1813665
Rather than selecting an arbitrary track to snap, we select the closest
track to our point, allowing the via to be placed along the full track
length.
Fixes: lp:1813324
* https://bugs.launchpad.net/kicad/+bug/1813324
Some microwave inductor lengths cause invalid outputs
for the calculations, which causes jagged outputs.
* If the request length is such that four arcs and no straight
segments is too long
* If the length is such that an N-turn coil is too short, but
an N+1-turn coil is too long. This can happen when the coil
count is small.
This patch doesn't fix the underlying geometric issue here - fixing
the first requires a numerical method, and fixing the second probably
needs an iterative approach. Both of these could benefit from
a refactor.
However, this patch does prevent the tools producing invalid outputs,
which can sometimes be quite subtle mistakes if the "jags" are small.
Fixes: lp:1792119
* https://bugs.launchpad.net/kicad/+bug/1792119
Jeff Young
Jon Evans
Andrei Pozolotin
Ian McInerney
Michael Kavanagh
jean-pierre charras
John Beard
Tomasz Włostowski
Simon Richter
Andrey Fedorushkov
MaPePeR
Seth Hillbrand
Wayne Stambaugh
Thomas Pointhuber