Re: looking for a maintainer (pcb-rnd) (mailing list links)
On Thu, 8 Dec 2016, Dima Kogan wrote:
Can you send some links of the mailing list discussion that started the
fork?
Meanwhile the archive is back up. I think the original announcement was
this:
http://www.delorie.com/archives/browse.cgi?p=geda-user/2013/09/01/02:49:59
(For some time I didn't really expect anyone else to use my fork but the
svn was public from early days).
This mail was in the middle of a big file format debate, on of the
discussions that happens from time to time ever since:
http://www.delorie.com/archives/browse.cgi?p=geda-user/2013/09/01
I think mainline has a branch for yaml now but it's not merged back to
master and master is still using the original .pcb file format as primary.
Pcb-rnd already switched to lihata as primary format (but keeps supporting
.pcb too). The .pcb flag-compatibility patch is not yet applied to
mainline - I think developers generally agree it's a good thing, the patch
is there, but it just doesn't happen in mainline.
So there's no single thread that triggered the fork. It's more like many
threads all ending up with a lot of discussion and no actual
conclusion or result/action. I can link a typical one:
http://www.delorie.com/archives/browse.cgi?p=geda-user/2012/12/08
The "Find rat lines" thread. More generally, how we indicate short
circuits. The original behavior is an UI misfeature. More or less everyone
agreed that it's not the right thing. In that thread we explored different
proposals to solve the situation. I had my proposal too, but I also tried
to collect and categorize all proposals on the list.
I did't know much about the code base back then so I didn't dare to
implement a feature alone. Instead I talked to one of the developers,
and we agreed that I would write the core of the functionality in a small,
stand-alone lib and he would interface it to the code. I wrote the lib
with test cases, but then in the general developer overload of the time
the interfacing didn't happen. About a year later I took the lib and made
the interface in pcb-rnd.
Looking at the svn log the interfacing took 1 week. And this was the
generic pattern that led to the fork: we spent at least that much time
discussing the feature on the mailing list, and mainline still doesn't
have it (or any alternative solution) to this day.
So in short, I just wanted to get these sort of things implemented, and it
seemed a fork was the best way to do it. Looking back from now, it was a
good decision: I think the farthest these features could get with mainline
would be unreleased git branches.
Regards,
Tibor Palinkas
Reply to: