gEDA-dev: Tighter integration of gEDA and PCB?

[Peter Clifton]

On Sun, 2007-06-03 at 09:05 -0600, John Doty wrote:
> On Jun 3, 2007, at 12:36 AM, terminator356 at users.sourceforge.net wrote:
> 
> > Also, what are your thoughts on eventual total integration of gEDA
> >  and PCB?
> 
> A large part of the utility of gEDA is that it *isn't* integrated,  
> but plays nicely with many different tools, PCB among them. The  
> biggest job I've done with gEDA is a VLSI chip design: no printed  
> circuit board or footprints there! I've also used gEDA in three  
> different printed circuit flows, but never with PCB: each customer  
> has their own favorite here. And then there's simulation.
> 
> So to me, it is important that gEDA remain neutral and not commit to  
> any specialized vision of the design flow.

When this "integration" comes, it will probably be in the form of a
framework which allows various different bits of glue between the
applications. Like gnetlist has various scheme backends for different
output formats, gschem could gain script based plugins for integration.
PCB, gnucap (and perhaps ngspice) are three possible examples I can
think of off the top of my head.

Related to this is intelligence about attributes. We should present
different recommended sets of attributes depending upon what workflow
the user is using, and appropriate context.

Past DJ's and Peter B's suggestions to syntax and utility, I'm still
stuck trying to figure out:

a) What the UI would look like. I've considered adding a number of named
"workflows" to the config file, which the user might pick from at some
point. This could be used to select which attributes are offered. When
(and how) should the user choose this? Could / should the a default list
of attributes offered for a component in a given work-flow be embedded
in the symbol?

b) How to specify / implement the attribute offering intelligence - what
the config file would be like?:

	- Do the definitions look like data which is parsed / used in a
	  particular way, based upon defined rules?

	- Are the definitions scheme code to implement the intelligence?


If anyone sees a nice (and ideally flexible) way to do this, can they
start some discussion here?

Regards,

Peter Clifton