Why I’ve stopped reporting bugs to Ubuntu

I’ve largely stopped reporting bugs to Ubuntu because of the condescending and dismissive attitude from their developers.

I cut my Linux teeth on RedHat back in 1998, and soon after settled on Debian as the best of several flawed but promising choices of Linux distributions. When I switched to Ubuntu 4.10 in October 2004, I was excited by the promise of a distribution with the quality of Debian and frequent releases and a focus on the desktop.

Today I stumbled across what unfortunately seems like another typical example of what happens when you report a bug to them: aumix in Ubuntu 7.10 was compiled wrong, such that it won’t even launch. Recompiling the source package without making any changes to the source fixes the problem. Instead of just doing that, the Ubuntu developers spent far more time and effort bickering on the bug report and justifying their inaction by referring to official protocol. Shallow thoughts outlines the issues with aumix and contains this quote:

When an organization gets to the point where it spends more energy on institutional processes for justifying not fixing something than on just fixing it — it’s over. –someone named Dave

The issue with aumix even elicited a post from Carla Schroeder at O’Reilly, and then an apology (unnecessary, in my opinion) from same.

Ubuntu needs to change its bug management culture. Here are some more of my experiences with them:

A bug in f-spot that went unaddressed for fifteen months, and was eventually expired by “Launchpad Janitor,” despite the fact that other people confirm that it is still happening a year and a half and four Ubuntu releases later. The only thing that Ubuntu did was ask me to confirm that it wasn’t a problem with my GTK theme. f-spot remains useless because of this bug.

A bizarre bug that was ignored for two months, only to have Timo Aaltonen declare, incorrectly, that the problem was in my fstab without ever asking to look at that file. The bug was closed without resolution, even though I replied stating that the Aaltonen’s explanation was incorrect. The bug silently vanished when I upgraded to Ubuntu 7.10.

My friend Jeremy reported this bug fourteen months ago. He included a patch, and then spent four months updating that patch and responding to Ubuntu’s concerns about it. Then, after nine months of inaction by Ubuntu, Martin Pitt rejected his fix and asked him to re-do it, because Ubuntu policy had changed. In his response, Jeremy points out the effect Ubuntu’s attitude has on him:

We keep our own apt repository of fixed packages and at this point I sometimes report bugs / upload patches and sometimes don’t. It doesn’t appear to be worth my time and effort to report / upload files. The reports are generally ignored and stagnate until the release is so old it’s mostly irrelevant.

And he explains how staggeringly unreasonable it was for Pitt to ask Jeremy to put in still more work on a patch that was now obsolete, especially when Pitt could have fixed it himself faster:

The fix you want is to a single word in the changelog version. It probably took the same amount of time to reject the package and post a comment as it would have taken to edit the patch and accept the package.

Jeremy’s complaint receives a pretty good apology from Emmet Hikory.

Then there were a few severe bugs I experienced while upgrading from 6.06 to 6.10. I reported these because the upgrade tool crashed with a message telling me to report them. The first went ignored for three months, and then Michael Vogt asked me to upload the /var/crash file, which (I assume) had long since been cleaned up by a reboot (or perhaps by me, removing a random, large crash report that nobody cared to see).

The second upgrade bug, however, takes the cake. I could not provide a stack trace: because the bug occurred during an upgrade, reproducing it would mean downgrading my entire OS, installing debugging packages, and re-running the entire upgrade inside a debugger. I also reported the bug to Gnome, and linked to that bug report from Launchpad. Sebastien Bacher immediately asked me to provide a stack trace, and two months later Daniel Holbach closed the bug because I hadn’t to provided one. Here’s how I responded to them on the bug:

I did not attempt to provide a backtrace since that would require downgrading to Ubuntu 6.06 (which would probably require a reinstall and possibly reformatting my disk), building debugging packages for gnome-terminal 6.06, and then re-running the upgrade process. That would take all day, or maybe multiple days, and I don’t have that kind of free time to have one of my two primary machines out of commission.

If you read the bug report that I submitted to Gnome, and linked from this bug report, you will see that I gave them on Jan 16th exactly the same explanation for why I wasn’t going to be able to obtain a backtrace.

In this context, Sebastian Bacher’s request on Jan 29th for me to obtain a backtrace seemed so ridiculous to me that I honestly thought it was generated by a robot, and nobody had actually read the details of my bug report. So I ignored it, waiting for a real person.

I don’t think either of these guys actually thought about my bug report before responding to it.

I understand that Ubuntu gets lots of bugs and must close out unresolved bugs that don’t have enough information. But the attitude on their forums is often a variation of blaming the messenger — either they blame the user for misconfiguring something (my theme or fstab), or ask them for more information that they cannot reasonably be expected to provide. And they often close bugs without really waiting to see if the problem has been solved.

The focus seems to be on reducing the number of bug reports and getting their users to do more of their work for them, rather than improving the quality of those bug reports, and actually understanding and fixing those bugs. I wouldn’t be surprised if Ubuntu is using some sort of performance metric internally, such as measuring QA performance in number of bug reports closed. Something seems to be encouraging their people to just close bugs as fast as possible.

Whatever the explanation, until Ubuntu’s bug management culture starts to change, people like me, who can actually help make Ubuntu better, will be less and less likely to contribute.

Update #1, 2008-01-18, 12:45 PST:

Less than a day after posting this, it’s received a bit of attention. Michael Vogt sent me this very gracious email, pointing out that one of my bugs was actually fixed:

Hi,

I read your blog post with some interest and I felt that I should
respond since you explicitly mention my name (I would have added a
comment to your blog, but that seems to be disabled).

I’m sorry that you feel that your bugreports are not handled well by
the developers. I would like to point out that the bug you reported
against update-notifier was taken seriously and got fixed for the next
release. Asking for a long gone crash file from my side was not
helpful, but I kind of hoped it might still be available.

Cheers,
Michael

Thanks, Michael. He’s right that asking for an old crash log couldn’t hurt, and I’m glad that that bug actually did get fixed. There’s a number of mostly reasonable comments over at OSNews. Sounds like a lot of people have been similarly frustrated. And two of my bug reports (in procps and f-spot) have had helpful, explanatory comments added. It would be great if this attention translates into an overall change in Ubuntu’s bug management culture.

Update #2, 2008-01-18, 14:00 PST:

Jeremy has posted Fixing and reporting Ubuntu bugs in response.

Update #3, 2009-01-18:

Ubuntu 8.10 is a bit better than 7.10 and 8.04. See also Patch: Ubuntu DRBD now can haz run after boot.