Re: gEDA: automatic netnames in gschem

[Carlos Nieves Ónega <cnieves-at-NOSPAMiespana_dot_es>]

El sáb, 10-12-2005 a las 16:07 -0500, Dan McMahill escribió:
[snip]
> > IIRC, the netnames could change if you use one backend or another, due
> > to netname restrictions in some programs...
> 
> This is certainly true that there are netname restrictions.  In fact the 
> place where that mapping is done (and the only place it is known) is in 
> the specific gnetlist backends.  It may not be the best way, but some 
> commercial tools deal with this by creating a database file at netlist 
> time which maps schematic names (node names, pin names for currents, 
> instance names, etc) to simulator names.  It would not be hard at all to 
> dump out this database because as part of the netname modifications, a 
> database is already created to detect netname collisions due to target 
> netlist name limitations.

Certainly this is how most commercial EDA work. 
I meant I'm not so sure how this scheme would work if a netname needs to
change between gschem and the other app. 

> > It may be not the best way, but a workaround for this could be the
> > following:
> > - Click on any net in gschem.
> > - Get the components and its pins connected to that net. For example:
> > pin 2 of R1, pin 3 of R2.
> 
> Does this capability exist within gschem already or is that part really 
> just a part of gnetlist?

If gnetlist can do it, gschem can also do it... I don't know if it can
do it already, but if not, it shouldn't be so difficult to add (just
look at how gnetlist does it).

> > - Tell pcb or whatever app to select the net connected to pin 2 of R1
> > and/or pin 3 of R2.
> > This way you handle nets as they are: a connection between pins of
> > components, and you are not tied to any name (it can change when you
> > netlist the schematic).
> 
> For pcb this doesn't sound bad, especially in light of DJ's reply to my 
> mail (that he had just added this to the hid sources).  I'm not sure how 
> well this may work with simulators.  I suppose since we have to create a 
> netlist anyway for simulation, that a database of sorts could be 
> constructed at the same time.

Let the simulator developers answer this...