DRC and split planes - what sorts of algorithms do programs use?

[peter-h]

I am using Protel PCB 2.8.

Yes it is ancient (1995) but it works, does all I need, and I know how to use it Also I often need to revisit very old projects - anything up to 20 years old. It runs fine in a winXP VM (VMWARE). I also have Protel 99SE which opens the v2.8 files apparently ok but so far have not felt the need to use it. I even found out how to generate gerbers for JLCPCB, who famously state they cannot read English and never read instructions on the layers, and who want apertures embedded within the gerber files.

However v2.8 can sometimes generate a lot of mysterious DRC errors. Some of these I have not solved and just leave them. Some are e.g. vias within SMT pads; it doesn't like that.

But others relate to "broken nets" where a plane has been split with a track, to get correct grounding routes etc. I don't like to ignore DRC errors; it is like ignoring compiler warnings.

Protel PCB requires any such breaks to be done with a polygon (a specific feature), not with just a normal track. But even when I do this correctly, it still reports some plane segments are being unconnected. This appears to be related to splits which have tracks at an angle, not 90 degrees, leading me to suspect that they use some sort of geometric algorithm to work out the connection topology and this breaks with polygons which have say 45 degree tracks. The user manual is silent on this.

OTOH I would have thought that algorithms for working out if you are "inside" a shape, for any arbitrary shape, have been thoroughly worked out in the 1960s, so why these issues in the 1990s?

I wonder if any modern PCB software does this sort of thing absolutely correctly?