Previously, (v7), the NPTH had a too-large clearance. This affected the
fill so that there was no island. In v8, at some point, NPTH had zero
clearance, which was definitely incorrect but also did not create an
island.
Correcting this to have the appropriate clearance between copper and
NPTH results in one almost island that has a neck and is appropriately
caught by the copper connection check.
The schematic net names are fully unique but if we have multiple pads
that are mapped to a NC pin, they need to have unique net names so that
they do not get connected to each other in the ratsnest. This breaks
the schematic parity check as we have modified the netname for some
pads. To work around this, we first maintain the zero-th net without
suffix and then add an additional check in DRC to ensure that we allow
unconnected nets as long as the share a common prefix with the schematic
netname.
We were calculating the same thing in three locations and we missed
adding the clearance from the footprints in, resulting in bad fills and
missed drc errors (see QA addition)
The RegressionZoneFillTests handle issues that are resolved by the zone
filler, so Clipper2 problematic fill belongs there. Otherwise, the
remaining tests should not refill the zones before running DRC because
we expect the DRC to catch errors without needing a refill
Adds QA checks to copper sliver tests. Adds the following checks:
- Dot product between two arms (quickly avoids checks for >90°)
- Checks the sliver is convex (area test)
- Eliminates minor slivers with angles that are approximately 0 and ones
with the opposite side width beneath a configurable level
- Updates Clipper2 to fix a couple of jagged edges on inflate
- Adds simplify during zone fill inflation to limit jaggies
Fixes https://gitlab.com/kicad/code/kicad/issues/14549
Jon Evans
Seth Hillbrand
Jeff Young
jean-pierre charras
Marek Roszko