This is another excellent test because it does some pretty weird stuff (such as rounded rectangle pads mimicking circular ones so that the thermal spokes are at 90º instead of 45º).
Anyway, I’ve pushed bits that make it mostly good now. There are still 4 unconnected thermal sinks where it was taking advantage of a bug which no longer exists (thermal spokes connecting to each other rather than to the zone). Cheers, Jeff. PS: the “fully” rounded rectangle pads exposed two older bugs: (a) the old thermal relief algorithm wasn’t taking the radius into account, and (b) the low-res/high-res optimisation needs to go to high-res for rounded rectangles with large radii (I chose above 40% as the cutoff). > On 24 Jun 2019, at 16:59, Seth Hillbrand <s...@hillbrand.org> wrote: > > On 2019-06-23 18:00, Jeff Young wrote: >> I’ve pushed bits to address Seth’s case. His case also had slightly >> different performance issues, so there are some more improvements >> there too. > > Thanks Jeff! That board works much better with the new code now. > > I'm attaching another corner case. This is the BeagleBone Blue project in > KiCad format. There are no DRC errors or unconnected nets in 5.1.2+. > > Best- > Seth<BeagleBone_Blue_KiCad.zip> _______________________________________________ Mailing list: https://launchpad.net/~kicad-developers Post to : kicad-developers@lists.launchpad.net Unsubscribe : https://launchpad.net/~kicad-developers More help : https://help.launchpad.net/ListHelp
