Why Ubuntu's bug tracking is useless

As a response to criticism from colleagues about flaws in Ubuntu I decided to make sure I reported and followed up every annoyance and usability flaw I found in Ubuntu. What I've discovered is that Ubuntu's bug handling system is so broken that it's most likely a net cost to the project rather than a gain.

First let's get the success out of the way.

In the hwtest program, designed to report back to Ubuntu working combinations, the UI is drainbramaged. You have to click a checkbox, then click ok for all the hardware you have. I reported this, and the author did get back to me and we discussed good solutions, which I believe may eventually get into a future release.

Now, for some random examples of where things break down.

If you change the pulseaudio settings rhythmbox locks up. After I reported this I was at least hand-held to produce a backtrace, although as far as I can see the bug never got pushed upstream. If it doesn't go upstream, my report is a waste of time.

Large window frames misrender: If you have a large screen (like I do) then the tops of the windows run out of paint before they get to the right side of the screen. Yes, this is cosmetic, but it looks ugly and is hard to resize windows (because you can't see where to click and end up hitting the close box).

Gnuplot can't plot log(x**0.5)/log(x) over the range 2 - 4. This is a very nice function, nothing tricky. But for some reason gnuplot just locks up. I rapidly got a confirmed meassage, but then it was marked as low and dropped on the floor. If it isn't forwarded upstream, why bother?

Indeed, with the exception of bugs which have been correctly upstreamed by Bryce and Mental, every bug I've submitted has been essentially ignored.

The problem is not with the triaging itself (people are very careful to check for the bug's validity), but rather in the general belief that all bugs which don't cause the triager's machine to catch fire and start world war 4 are either wishlist or low priority. It's perfectly valid for a project to have a high priority bug which is simply not able to be fixed. Marking a bug as low priority simply to give the impression of good quality is disingenuous, pisses off the small concientious part of the community who try to support Ubuntu by providing usability bug reports, leads to a bad user experience for others and dismisses good starter projects for newbies as unimportant.

I believe that the most important thing for Ubuntu to do is make its bug reporters feel important, and furthermore, to provide accurate reporting for free software projects. If a project can not get good bug reports, especially of things which upset the end user, it will never amount to much. Without commitment to the low end bugs Ubuntu is never going to compete with MSVista, let alone MacOS.

Full disclosure: Now to be fair, I've had one bug assigned to me which I haven't resolved (surface area and contour length in inkscape). Well the contour length is done (and was done 4 years ago), but surface area requires exact unwinding, which is not yet robustly implemented in 2geom. So until then it's not fixable.

Comments

The gnuplot bug was upstreamed. (02:07:05 PM) mental: fwiw, they have recently set up a thing which gathers metrics on bugs getting upstreamed. There was a presentation on it at UDS with the emphasis that not enough is getting upstreamed yet. (08:57:26 PM) peter: I don't agree with "Without commitment to the low end bugs Ubuntu is never going to compete with MSVista, let alone MacOS.", for a couple of reasons: (i) although I don't use vista and there's significant sampling bias in the people whose opinions on the matter i've heard, I do get the impression that Ubuntu is already competitive with (i.e. roughly comparable with) MSVista. Windows-centric "PC Magazine" here in .au has for some months had Gnu/Linux-related content on their front cover, starting (to my knowledge) with a headline along the lines "Vista Killer" a few months ago. (A single headline can be put down to sensationalist journalism, but the fact that they've followed up with content from then on suggests that they seriously believe it's of interest to general PC users.)

In any case, evidently at least a very small number of people have found free software better for their needs than Windows for over a decade.

(ii) Free software can be competitive even in cases where it's more buggy than the proprietary competition.

Those objections made, I will say that one of the strengths of free software is that it does allow users to fix bugs, and certain types of bugs are more likely to be addressed in free software than in proprietary software. I.e. I think one of free software's advantages over proprietary programs is something to do with bug fixing, and that mishandling of bug reports erodes such an advantage.




[æ]