kicad-developers team mailing list archive

[Dick Hollenbeck <dick@...>]

Jean-Pierre,

The attached board file is a modification of kicad's demo 
"sonde-xilinx.brd", and then fed through the specctra export and then 
imported back in. (So the interface is now largely done and I am mostly 
going back to work on my boards.)

However, if you load this attached board and click on the ratsnest 
button, you will see that freerouter is routing tracks to the center of 
the "oval pads with offset", not to the through hole center. So the 
ratsnest is catching this and thinking it is an unrouted pin/pad.

I am not inclined to work on this particular issue, but I wanted to 
point it out in case you wanted to try and change what we consider the 
"connection point" within the pad. (We have had a discussion like this 
before. It is very confusing for a user to see a track go to a pad 
visually and then have it not be recognized as "connected" by Kicad. 
Maybe other software does this too?)

If we were to change what the "connection point" is (and still make it 
be a "point on point" rather than a more tolerant overlap), we would 
probably have to allow either the old connection point or the new 
connection point when ratsnest testing of "oval pads with offsets".


Please give this some thought. I see 3 courses of action we could take: 

1) none.

2) allow two connection points for "oval pads with offset" only.

3) become tolerant of track on pad overlaps generally according to some 
more tolerant but not so execution expensive algorithm.


FYI,

Dick