Mail Archives: geda-user/2012/11/18/06:42:44
On Nov 17, 2012, at 10:17 PM, Evan Foss wrote:
> From my view point the rest of gEDA/PCB should change around gschem to
> better use it and not the other way around. There is a lot of unused
> metadata in a schematic that is yet unused and that is where most of
> my frustrations with gEDA come from. This is why see issues with the
> netlister. Following the Unix philosophy it does it's job providing a
> netlist but that removes a lot of the schematics metadata that other
> tools like PCB might want. For example which connections are a bus or
> meant to be routed differentially? There are comments I leave on
> schematics in gschem that I would like to be passed to PCB tools that
> get removed because we only deal with netlists. I have been
> contemplating for a while now writing something to be parallel in the
> work flow with gnetlist to just handle this stuff. I view this as
> working from gschem at the bottom up which is the opposite of the way
> modern top down development process goes but I am curious to see an
> opinion from the group.
There even exists such a thing, in production use by me, anyway.
https://github.com/xcthulhu/lambda-geda
It's gnetlist-like in that it extracts the data from the schematics and presents it to a "back end" script. But unlike gnetlist, the back end has access to all of the data, and control over its processing. So far, the only useful back end invokes hierarchy expansion, and then outputs the flattened schematics.
Unfortunately, it's written in Haskell, which effectively means that only its author can write back ends for it. He has expressed an interest in rewriting it in Guile, which would put it more in the gEDA main stream, although it seems only a small minority of us can write Guile scripts.
John Doty Noqsi Aerospace, Ltd.
http://www.noqsi.com/
jpd AT noqsi DOT com
- Raw text -