gEDA-dev: refdes_renum

[Stuart Brorson]

On Mon, 12 Feb 2007, john at ecosensory.com wrote:
> Dan McMahill wrote:
>  The nice thing about
>> guile (if it has been incorporated in a way that provides a stable api) is 
>> that it is insensitive to any file format changes.  refdes_renum is not.
>> 
>> In general I'm in favor of expanding what can be done in guile because it 
>> is pretty useful to have a highly capable api for manipulating a CAD 
>> database.
>
>
> Hear hear,  Guile/scheme/lisp  is jsut right for dealing with the lists of 
> attributes we have attached to drawing objects in PCB.
>
> John Griessen

I'll just chime in with my $0.02 here.  Then I'll slink away again.

IMO, it's nice to have multiple ways to number the refdeses.
Certainly, using guile to autonumber the refdeses in gschem is a "good
thing" (TM), and I support any enhancements to that tool, including
the two  features which were suggested during the code sprint, i.e.:

1.  Be able to clear all refdeses (e.g. C57 -> C?)
2.  "Gently" renumber refdeses.

That being said, I also like -- and tend to personally use -- Dan's
refdes_renum tool.  Here are some reasons to keep it alive, and
indeed continue to nurture it:

*  More people feel comfortable hacking Perl than hacking guile &
binding it into gschem.  Therefore, a Perl-based tool has a larger
support base.  For example, John Luciani implemented feature #1 last
Saturday.

*  I (personally) like having an external refdes tool, particularly if
it can do many more things than the gschem's built-in guile tool.  I'd
probably end up using both, depending upon situation.

*  Multiple tools performing one task fits well into the Linux way.
It allows for cross-pollination, friendly competition, borrowing
ideas, and so on. Therefore, there's room for refdes_renum, grenum,
and so on.

*  As for file format changes, you do have a point, but I think the
risk that something will break REFDES=C? is low, and fixing the Perl
in case it does break is easy.

Finally, this thread started because Rikster wanted to hack
refdes_renum.  Since C56 -> C? is done already, maybe he'll be
interested in implementing a --gentle option for refdes_renum?  But
he'd better hurry, since I'm getting itchy to do it myself!  :-)

Stuart