Karl:
I'm not "complaining" about "other people" to you. I was bringing them up
as something to understand in the emotional climate that I am in as I typed
the previous e-mail. I do find it interesting that you don't care much
about the fact that "other people" have complaints about Subversion.
Touche on your "are you volunteering" response. Actually, I _would_ be
willing to rip out all of the read only file code if this hasn't been done
already (which would be "solve [the] problem[]" on "the code should be
rolled back"). Brane asked if this could be done a while back and I believe
the discussion died for some reason (which was interesting, as I remember
Greg being mad at the read only code himself for other reasons).
Alas, I am unable to do much about people who add files to the project and
don't get them into the Windows project files. The fact that people aren't
helping other developers and notifying them when they add files is the
problem here, not neccessarily something that a third party can efficiently
address. This is a policy issue on commiters, and a good example of the
kinds of things that would help non-Unix developers.
If you want something I _could_ offer (and likely would), it is either to
futz with the build enviroment and build an automatic dsp generator (which
is likely to be important as the APR people got so mad when I... nicely I
might add... asked if VC++ 7 could be supported and they yelled at me that
VC6 might not even be continued to be supported until someone wrote an
automatic project generator). I am likely to be doing this for my own
purposes anyway, although I think that someone else was already volunteering
to work on such a solution for APR, which would make the effort quite
duplicious.
The patches I sent in weren't ignored... the patches I sent in _were_
merged, and a good amount of my time was put into trying to deal with these
various issues. It was more that they weren't enough to deal with the
issues (I simply didn't understand how Subversion worked well enough to
address most of them), so the issues were sluffed off as unimportant and
mostly still left unfixed. The point of the sentence had nothing to do with
the patches themselves (and maybe the effect would have been better had I
not even used the word "my", I had a feeling someone would nit on that), but
on the attitude that went into such a milestone release.
Tasking outside developers, who by nature aren't at expert level at how
Subversion works, exclusively to fix issues that seem core to the
abstractions (such as case sensitivity and the occasionally weird behavior
of directory seperators) is not a sufficient way to deal with these problems
(a point helped in evidence of this project), and I hate to break this to
you, but don't require much platform specific knowledge at all. You just
have to know some of the limitations of the other platform (such as "case
insensitive" and "uses \ instead of /", we aren't talking things that are
that complicated) as you develop. If you want other people, including me,
to be able to fix these issues, you have to devote more of _your_ time to
understanding the issues, finding the problems, and yes, even thinking a
little bit like a Win32 programmer.
To point out, similar issues to the ones I'm bringing up have come up in the
past in a thread called "Subversion portability", started by Brane after a
"sermon" he wrote to another thread. The entire thing is rather well summed
up by the first e-mail to "Subversion portability" that contains both the
sermon, a response, and a few extra points. The arguments are quite similar
to mine, and only slightly less inflamatory :).
Sincerely,
Jay Freeman (saurik)
saurik@saurik.com
----- Original Message -----
From: "Karl Fogel" <kfogel@newton.ch.collab.net>
To: "Jay Freeman (saurik)" <saurik@saurik.com>
Cc: "svn-dev" <dev@subversion.tigris.org>
Sent: Wednesday, June 05, 2002 3:48 PM
Subject: Re: GUI for subversion?
> "Jay Freeman \(saurik\)" <saurik@saurik.com> writes:
> > I'd more like the Unix people to be careful about the Windows builds.
When
> > new files are added, people should be notified so the build scripts can
be
> > updated. When more fundamental issues come up (such as the major
read-only
> > regression that likely still hasn't been fixed), the code should be
rolled
> > back. There was even a milestone that almost went out a few months ago
with
> > a Win32 client that couldn't even do a checkout. There was more
interest in
> > getting the milestone out for the Unix people than getting the patch
that I
> > had sent in applied to the Win32 build. I am pretty sure it got
applied,
> > but only because I made an annoying argument to the case.
>
> I notice a lot of passive voice here. "People should be notified",
> "The code should be rolled back", ...
>
> Are you volunteering to solve these problems?
>
> If you're upset because you sent in patches which got ignored (which
> is justifiable), please post complaining about that issue. When you
> do work, like making a patch, we should appreciate and take advantage
> of this, that's for sure.
>
> But if you're proposing that people who have neither the experience
> nor the equipment to maintain Win32 code should do so anyway, that
> just doesn't seem realistic to me.
>
> Subversion works well on Unix because more Unix developers
> participate. It works on Win32 to the degree that Win32 developers
> participate (like Brane, et al, bless 'em). No one's
>
> You talk like no one has ever caused a Unix regression, like there's
> some conspiracy to frustrate Win32 developers specifically. There
> isn't. Subversion is just like any other project -- it works where
> people test it. If you want it to work better under Windows, please
> help make that happen.
>
> > I understand that my e-mail here is a little abrasive, but it's a
> > culmination of being fed up for months, dealing with other people who
are
> > now thouroughly convinced that we won't be able to switch to Subversion
for
> > the projects I have been advocating it for due to the seeming lack of
Win32
> > empathy, and a number of mis-starts on an e-mail that voices these very
> > sentiments. I finally gave up when I noticed that a response to the GUI
> > message started turning into that same thread in my head and just wrote
it
> > out.
>
> No one here asked you to deal with those "other people", so don't
> transfer their complaints to this list.
>
> -Karl
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Jun 5 23:18:53 2002