X-Authentication-Warning: delorie.com: mail set sender to geda-user-bounces using -f
X-Recipient: geda-user AT delorie DOT com
Date: Sat, 12 Sep 2015 19:01:02 GMT
From: falcon AT ivan DOT Harhan DOT ORG (Spacefalcon the Outlaw)
Message-Id: <1509121901.AA26523@ivan.Harhan.ORG>
To: geda-user AT delorie DOT com
Subject: [geda-user] A call for a separate PCB mailing list
Reply-To: geda-user AT delorie DOT com

Hello fellow adherents of free EDA tools,

Someone made a suggestion to create a separate mailing list for PCB,
and I would like to second that suggestion.

Just like John Doty, I cringe at seeing gschem and PCB being lumped
together, but for a completely different reason.  My perspective is
the opposite of John's: whereas John does graphical schematics in
gschem and is fortunate enough to have someone else do PCB layout for
him for a price he can afford, I don't use graphical schematics at all
(see my previous post), hence I don't need gschem.  But I am not rich
enough to afford San Diego PCB's prices for PCB layout labor, hence I
need to make other arrangements to get my layouts done, arrangements
which require that the PCB layout step be done with free software
tools, no proprietary software allowed.  Hence I am highly interested
in FLOSS PCB, but not in gschem or other schematic capture programs.

For the next month or two I will probably take a break from sw/fw/hw
hacking to work on my electrolysis (long story, too far OT for this
list), but when I come back, I plan to start working in the earnest on
a "next generation PCB" that will do what I need.  I have a need to
build cellular phones and modems with the Calypso chipset, and this
need won't go away.  Being much less integrated than the newer ones,
this old chipset imposes high demands on PCB layout: discrete GHz RF
components which need GHz RF trace interconnections on the PCB, and
lots of digital signals to be wired between chips in a very tightly
squeezed layout.

The historical designs I follow were done in PADS, and use features
which GNU PCB or gEDA/PCB or whatever it should be called currently
lacks.  Having found the docs for PADS' ASCII PCB format, I would like
to write an automated lossless translator from PADS to FLOSS PCB which
would allow me to reuse a Calypso GSM modem design from PADS without
losing any of it and without having to redraw any of it manually, but
one cannot meaningfully talk about a translator from A to B unless B
supports all functionality of A, or at least all of the subset of that
functionality which is used by the design to be translated.  Thus
before I can start working on a translator from PADS to FLOSS PCB, I
first need to develop a "next generation" FLOSS PCB that matches PADS
in capabilities, at least for the subset used by the GSM designs of
interest to me.

When I come back from working on my electrolysis (more precisely, from
studying it and experimenting with it together with my significant
other until I get her to the point where she can do electrolysis on my
face), I will start working in the earnest on this Next Generation PCB.
You can be quite certain that this work will happen indeed, as I need
this NGPCB in order to build my Calypso GSM boards, and I really,
really, really want to build those boards.

But what I am still undecided on is whether or not I will make any
attempt at integrating my NGPCB work with the present community, or if
I would rather do it as a completely divergent fork.  As a fork, it
would probably be even more divergent than Igor2's pcb-rnd, as I would
need to change the PCB data model to be more PADS-like, which in turn
would entail a significant non-backward-compatible file format change.

As I won't be actually starting this work until after I come back from
my electrolysis sabbatical, I have a couple of months before I have to
make a decision on staying with this community or forking.  But when I
do come to the point of having to make that decision, a major deciding
factor for me will be the mailing list situation.  I don't think I'll
be able to do my NGPCB work with mainline integration in mind unless
PCB gets its own mailing list that is completely separate from gEDA/gaf.

The Next Generation PCB project will be a lot of work, and there is
simply no way I could do work of that intensity while simultaneously
having to read a hundred emails a day endlessly arguing about
integration vs. separation between gschem and pcb, and other such
issues of no relevance to a PCB-only user.

But given how the vast majority of people on this list use gschem and
pcb together, and seeming that there are only two people in the whole
world who use just one of the tools (John Doty uses only gEDA/gaf; I
am the opposite), I see there being very little realistic chance of a
separate gaf-free mailing list being created for PCB.  Therefore, I
will probably end up hosting my NGPCB project on freecalypso.org as a
fork with its own separate source repository and mailing list.

SF