Re: Lots of problems with upstream, please help!
-----BEGIN PGP SIGNED MESSAGE-----
Le 23/10/2012 06:38, David Smith a écrit :
> Hello, I hope this is the right mailing list.
> I've been a Debian user for over 10 years, but due to schoolwork
> and work i haven't been very involved in Debian. Mostly I have
> submitted patches to upstream or helping upstream debug rather than
> helping Debian specifically.
When you help upstream, you do even more than helping Debian.
> For a long time, I always end up with a dozen or so of my own
> custom packages which I derive from official debian source packages
> with an extra tiny bugfix patch or two I've thrown in on top and
> recompiled. Sounds pretty selfish, I know.
Well, now you may have the skills to submit your patches :-)
> With Wheezy coming out, I'm trying to help any way I can but I'm
> having some problems.
If they're release critical, please do so. Else, just prepare Wheezy+1
:-) When Wheezy is out, you can *backport* those features in
As a general remark on all your problems, you have already realized
something very important about how free software development (does
not) work. Actually, this is not limited to free software (like, at
all): nothing worth having comes easy. Very often it takes months of
prodding the right person before your patch is accepted.
> Technically they are all problems with upstream and they're all
> for different software. I really don't know how to deal with them
> to help Debian.
> 1. As a big regression from Squeeze, I had to blacklist an HP
> platform driver from loading into the kernel in order for my
> wireless to work well. I filed a debian bug report about it and the
> debian maintainers helped me collect all the information I needed
> and submit it to upstream. I did all that, it's been 3 weeks, and
> the upstream kernel platform devs seem to have completely ignored
> it, although linux-wireless confirmed the platform driver is doing
> something very strange. So I don't know what to do. Maybe I just
> need to wait longer?
3 weeks is short. Is the bug release-critical?
> 2. About 8 months ago, an upstream developer wrote a patch to
> enable a feature in upstream software. Upstream didn't review the
> patch so it never officially made it into upstream software.
> Requests to review the patch by me and others are getting
> completely ignored upstream. The upstream dev that made the patch
> looks to have disappeared. Upstream seems reluctant to accept,
> reluctant to reject and reluctant to review, so it stays in limbo.
> Without upstream's proper review / approval of the patch, in
> addition to the fact that the patch essentially enables a feature,
> it's probably unlikely to ever make it into Wheezy, right? I filed
> a bug report saying that the feature didn't work as the feature is
> listed in the software as being available, but it's not actually
> enabled in the code. I'd imagine an NMU for enabling this feature
> in Wheezy would be inappropriate and the proper Debian thing to do
> would be to just write a different patch to remove the feature so
> the user doesn't see it in the software? Although, that's pretty
> much the exact opposite of what I'm doing on my PC so I really
> don't have much motivation to do that.
Unless the feature is central to the software, this doesn't sound
release-critical and even a patch to remove any trace of the ghost
feature would be inappropriate. On the other hand, if you can convince
the release team, the upstream patch could perhaps directly be
included in the Debian package. Or you can fix it post Wheezy and
provide a backport.
> 3. A couple months ago, an upstream developer wrote a patch to fix
> what I consider to be a serious bug that makes a lot of apps
> unusable under normal (aka "user customized") conditions. Upstream
> reviewed and approved the patch but the patch is written for KDE
> 4.9. 1. I tried applying the patch to KDE 4.8. 4 in Wheezy which
> ended up as a spectacular failure during compile. I asked upstream
> to back port the patch to kde 4.8. 4 in Debian and they said that
> they do not control the release of Debian and that KDE 4.8.5 is
> available. I'm almost certain the patch upstream wrote isn't
> going to work with 4.8.5 either, but I haven't tried it as I don't
> want to be that far away from Debian's packages. There is a
> work-around that's very easy to do, but it's not obvious, and it
> saps performance from my PC. I'm certain some Debian users running
> KDE will eventually run into this very annoying problem and have to
> Google and enable the work around. I will likely end up backporting
> the patch myself for my own personal needs if the workaround annoys
> me enough some day, which it probably will.... eventually... I
> worry that if I backport this patch, it might be a little ugly and
> probably not suitable for Debian anyway, although I really haven't
> looked at backporting the patch yet.
Bug report number please? If the conditions under which the bug bites
are really "normal" as in statistically significant, you have a
release-critical bug here. If you can prepare a patch, even ugly, it
will help the package maintainer to decide whether and how to fix this
bug for Wheezy.
> 4. There appears to be a regression in some streaming libraries.
> When I'm running wheezy and I replace some audio streaming
> libraries in Wheezy with the ones from Squeeze, it fixes the same
> problem in several different audio players in Wheezy. What's very
> strange is that some other audio players don't have the problem at
> all in Wheezy with both the older and newer libs. I want to say
> that it's all the audio streaming libraries fault, but at the same
> time it appears that the recommended way for audio players to
> interface with these audio streaming libraries may have changed.
> It's so very complicated and I'm having a lot of trouble debugging
> since it's not crashing ( I wish!) and only periodically throwing
> out a warning or two in the background. [...]
A warning or two really is not a matter. But if there is an actual
failure, by all means please send a bugreport to whatever package you
guess is responsible. You don't have to fix a bug before you may
submit a bug report!
> Really, I'm having a lot of problems with upstream and I'd like
> some help/suggestions on the "right" thing to do to help Debian.
> For the most part, I have been creating bug reports in both Debian
> and upstream and "forwarding" the debian bug report to upstream so
> Debian maintainers are aware of the problem.. But I'd like to do
- continue doing what you're doing
- write patches, even if they are ugly. That will help the maintainer
understand what the problem is and may help hime write a better patch.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
-----END PGP SIGNATURE-----