Re: gEDA-user: autorouter fixes and enhancements (Harry Eaton)

[Anthony Blake <tonyb33@xxxxxxxxx>]

Kai-Martin Knaak wrote:
> [...] Any hints that let me actually use the router appreciated.

Uhh, I don't think you could actually use it for anything useful, as 
it's pretty broken.. but if you still want to go ahead try 
:toporouter(). There is a 'h' parameter which makes it re-evaluate the 
netordering after a number of routes, so :toporouter(h20) would make it 
recompute a net ordering after 20 routes. To keep it out of a 
layergroup, use 'l' followed by the layergroup number.

There are a few major problems which will be sorted out in the next commit:

I have a whole lot of really bad code to try and make the layout more 
aesthetic by forcing traces to leave pads orthogonal to the edge. This 
messes with some of the clearance code and can sometimes result in 
impossible arcs. For example, to get the LED board to work, you must 
change the square pin of J3 to a round one. I'm removing this code.

The way the curvilinear wiring is computed is overly complicated and 
breaks in all sorts of situations (lots of slithery triangles shows it 
up quite nicely, e.g., under a TQFP). In retrospect I'm embarrassed I 
tried to do it that way. I reckon the new code will shave over 2k lines 
too, its *way* simpler.

There is a whole lot of dodgey clearance code duplicated all over the 
place, which gets messed up by the situation where another route is on 
both sides of some other route simultaneously (where a hairpin occurs), 
and also gets messed up by the special cased arcs coming out of pads. 
This is all being replaced by one set of functions which actually work.

Good luck with that..

-Anthony



_______________________________________________
geda-user mailing list
geda-user@xxxxxxxxxxxxxx
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user