[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: gEDA-user: Strange problems in multi part symbols

On Jan 2, 2011, at 6:14 PM, Oliver King-Smith wrote:

>  Likewise spice back end also blows a
>   gasket trying with the pinseq numbers.  I thought the pin seq refers to
>   which position in the spice parameter list the pin refers to?
>   I was going to see if I could fix up the spice-sdb backend to handle
>   putting multi part symbols back together, but I think something might
>   be more fundamentally broken inside gschem.  Do folks have any
>   thoughts?

pinseq is overloaded. It is impossible to use slotted symbols for simulation. Stuart tried mightily to fix this and failed.

It could be fixed if the slot semantics were moved out of the core code. Then we could have the standard helpers in gnetlist.scm collect all slots for a given refdes into one package (for printed circuit flows) but simulation flows could separate slots into separate packages. This requires removing the special semantics for refdes from the core: the "same refdes, same package" rule is only appropriate for printed circuits.

Given that Stuart understood gschem and gnetlist better than most and put a lot of effort into fixing this, I believe that this cannot be fixed without more significant refactoring than he was willing to attempt. I don't think there's anything you can do in Guile alone to fix this.

John Doty              Noqsi Aerospace, Ltd.

geda-user mailing list