A couple of rants about Launchpad

[Colin Watson]

On Fri, Mar 06, 2009 at 09:48:32AM +0800, Christopher Chan wrote:
> Colin Watson wrote:
> > My bug triage comments have very little to do with fix rates, actually.
> > It's a bit of a disingenuous leap to go from "there are these specific
> > problems with bug triage in Ubuntu" to "nobody ever fixes bugs filed in
> > Launchpad"! Brian Murray recently blogged
> > http://qa.ubuntu.com/reports/bug-fixing/jaunty-fixes-report.html, which
> > is a list of all bugs that have been closed by package uploads
> > throughout the Jaunty cycle so far. Note that those are only bugs that
> > were explicitly annotated in a package changelog; it does sometimes
> > happen that you fix a bug just before somebody reports it and then you
> > close the bug separately, and that kind of case won't show up in Brian's
> > report. Nevertheless, people who say that we do not fix bugs filed in
> > Launchpad are simply misrepresenting the truth, knowingly or otherwise.
> 
> I am referring to rather embarrassing and obvious bugs like Network 
> Manager in Gnome that required expert intervention in releases. 
> Installing Ubuntu 7.04/7.10 but then having to talk my then colleague 
> through the terminal on the phone to fix the problem does not leave a 
> very good image of Ubuntu.
> 
> Stuff like that should have been nailed in the Release Candidates and 
> never released.

I don't remember the details, so can't comment on them specifically. We
spend the last couple of weeks of a release cycle doing very little else
but testing, and we try very hard to make sure that problems encountered
during testing are dealt with where it's at all possible to do so
(tracking them via http://iso.qa.ubuntu.com/ and the iso-testing tag on
bugs). Of course, it does happen that problems slip through that affect
only a subset of people; sometimes they actually didn't come up during
testing at all, and sometimes they did but we just weren't able to fix
them in time and had to resort to workarounds to minimise their effects.

When this sort of thing hits you, of course it's natural to say "oh,
those Ubuntu folks, they couldn't be bothered to test and fix things". I
don't blame you for that. I'm simply here to say that that is not the
way it works. We make frequent calls for more involvement in pre-release
testing so that it isn't just a small number of regular testers
involved.

> Besides bugs, there are other issues. I had eight computers I was hoping 
> to use as a pilot. Well, now I am down to one machine because Debian 
> Installer does not have sufficient documentation to tell me how to do 
> automatically do raid and lvm besides having to familiarize myself with 
> what looks like hundreds of variables. Automated tools for installation 
> and package management are very important for big deployments and I am 
> now seriously rethinking my adversity to Fedora since things in Ubuntu 
> land seem pretty much the same. There goes my hope of putting Ubuntu on 
> the desktops instead of Windows...and Promethean just released 
> ActivInspire on Linux too although they have missed 64-bit packages. 
> Better documentation would have helped me a long way with the roll out 
> of the pilot. Hitting the lists offered me no help at all. The best 
> response I got was that it was possible but I got no reply when I asked 
> for documentation on how to get it done.

I won't take responsibility for every single mailing list post being
answered correctly, but you should have been pointed to the installation
guide (https://help.ubuntu.com/8.10/installation-guide/i386/) whose
preseeding appendix does document how to do LVM or RAID partitioning.
That said, LVM *and* RAID partitioning (i.e. LVM on top of RAID) is a
common request that the automatic partitioning software can't quite
handle yet. I'm getting hammered on this from a number of directions, so
there's likely to be work on this for 9.10.

For those who find that preseeding is too complicated, Kickstart
compatibility is available
(https://help.ubuntu.com/8.10/installation-guide/i386/automatic-install.html).
It isn't as flexible in various areas, but that's OK since it provides
an escape hatch so that you can mix and match Kickstart and preseeding.

> > As Matt Zimmerman said a while back
> > (http://mdzlog.wordpress.com/2008/10/29/ubuntu-quality/):
> >
> >   When 8.10 is released, as with each previous release, some users will
> >   be disappointed that it has a bug which affects them.  This is
> >   regrettable, and I feel badly for affected users each time that I read
> >   about this, but it is unlikely to ever change.  There will never be a
> >   release of Ubuntu which is entirely free of bugs, and every
> >   non-trivial bug is important to someone.
> >
> > In the real world, anybody who tells you that their product is free of
> > bugs (or even free of significant bugs that you're going to care about)
> > is selling something.
> 
> You can at least ensure that certain basic configuration tools work 
> before you put them in a release. It gives a very bad image when I have 
> to walk a person through the terminal running commands to work around a 
> bug in a configuration tool for a basic need.

As above, we do test all sorts of things including configuration tools
and make sure they work. Obviously when something slips through for some
subset of people the natural conclusion for them is that we didn't
bother testing configuration tools at all, but that isn't the case. Our
testing does have holes and we're always trying to improve it, but it is
not absent by any means.

> There is another thing, though not really Ubuntu specific, that would
> be great is getting kiosktool ported over to KDE 4.2.

Not something I know anything about myself, I'm afraid ...

-- 
Colin Watson                                       [cjwatson at ubuntu.com]