kicad-developers team mailing list archive

[jean-pierre charras <jp.charras@xxxxxxxxxx>]

Le 21/04/2012 00:36, Dan Chianucci a écrit :
> Jean,
> I thought about this, and I agree with your first point.  This patch has shortcomings in the fact that things like wire stubs and
> unwanted lone tracks will also have there net-codes updated and in turn might not be caught by drc.
>
> I'll look into connecting the vias directly to a zone instead.


It is just an other approach, I am not sure it is better, but it has serious advantages.
It fixes an issue, and could also creates others (pay a particular attention to the fact
zones after edition can be merged or split, or can have holes).

Be sure fixing this floating tracks issue is very tricky, regardless of the approach.

An enhancement to the current patch could be:
- after recalculating net codes, test if tracks flagged floating are inside a zone outline:
ends inside a zone outline, same net and same layer (or connected to a via/track already inside the zone),
and set netcodes to 0 for really unconnected tracks.

In all cases keep in mind tracks and zones can be modified,
and detect floating tracks only when recalculating net codes could create issues after track/zone edition
and before the net codes are recalculated.

The problem to detect floating copper areas still exists,
and it is also tricky with floating tracks/vias.


> Dick,
> There are a few bug reports related to this, specifically
>
> https://bugs.launchpad.net/kicad/+bug/593923
> and
> https://bugs.launchpad.net/kicad/+bug/612133  (im not sure if this is the same thing or not)
>
> -Dan C

These bug are not related to zone filling.
The first was created by a bug in connection calculation algorithm:
under certain circumstances, connections were missed.

I rewrote it some times ago, and the new algorithm is by far more powerful.
( 10 times faster, and better detection of tracks to pads/tracks and pads to pads connections )

--
Jean-Pierre CHARRAS