Thursday, April 21, 2011

Ubuntu Bug Reporting 'Linux Journal'


Nov 04, 2010
Canonical does some things quite well.  Obviously, or Ubuntu would not have captured approximately 57% of the Linux desktop market share (according to Google's search insights tool). But there is at least one area for some major improvement:  bug reporting/tracking.  
In a previous LJ article I noted that an apparent long term bug had again resurfaced in Ubuntu 10.10.  This is a problem in the driver for some of the Atheros wireless chipsets which essentially renders those machines unusable.  What I did not discuss in that article was the incredible frustration involved with attempting to submit a bug report to the Ubuntu developers.  I’ll describe that process here.

Let’s say that you have discovered a problem in Ubuntu:  what do you do?  Well, I went to http://www.ubuntu.com/ to look for their bug reporting system.  So far, so good, there is a “Get support” link right in the middle of the page.  Click.  Now I’m on a page that offers to point me in the direction of free support from on-line forums, or to sell me “professional support services”.  Fair enough, everybody expects Canonical should be able to make a Krugerrand or two out of their distro business.  Down at the bottom of the page under the “Community” section there is a “Report a problem” link.

Click.

We are now at a page that describes how to report a problem, including a link to abug reporting tutorial.  Click.  Aha! Now we are on a page which actually has a link to Ubuntu’s launchpad.net bug reporting site.  This tutorial page suggests that you search for any existing reports of your bug before submitting a new report.  So far, so good still, if a bit convoluted.

Now, as a reminder of what led us here, a few days ago I had found a bug in the driver for the Atheros AR9285 wireless chipset in my Acer Aspire One N450 netbook.  The bug being that the netbook would not stay connected via wireless for more than a minute at a time.  So I clicked on the search link.  Down at the bottom of my screen appeared the message “Waiting on launchpad.net...”  Get used to that, you will be seeing it a lot if you are trying to research or report a bug.  Their launchpad bug reporting site is horribly slow.  You’ll also be seeing this message more often than you would probably care for:

Timeout error

Sorry, something just went wrong in Launchpad.
We’ve recorded what happened, and we’ll fix it as soon as possible. Apologies for the inconvenience.
Trying again in a couple of minutes might work.
(Error ID: OOPS-1752E1761)
In fact, this error message was all I was able to get out of launchpad when trying to do a search on “Atheros AR9258” using the “newest first” search option.  The site is almost always broken when you try to use that search option.  Finally after 6 tries the site returned 21 bug reports, many of which described a problem very similar to what I had observed, i.e. the wireless connection was useless with certain Atheros chipsets. Doing a more general search on "Atheros" "newest first" (only three launchpad error timeouts) returned 189 results.

So, just to be certain that Canonical was aware of the problem in their brand new Maverick Meerkat 10.10 release, I filed my own bug report: bug number 660864, if you want launchpad to find it for you.  A bit of additional research brought to light the fact that this problem with the Atheros drivers has been a recurrent one in Ubuntu dating all the way back to 7.04.  It certainly did not exist in the previous 10.04 Ubuntu Netbook Edition. Well, it did exist in the "out of the box" install, but installing the linux-backports-modules fixed the problem.
This experience leads me to conclude that Canonical has two significant areas for improvement:

  1. They could provide a less byzantine, convoluted, and painfully slow pathway to bug reporting.
  2. They should strengthen their regression testing procedures to help prevent previous bugs from re-entering their distributions.
Don’t misunderstand me, I like Ubuntu -- I’ve used it for years.  However, it is definitely time for Canonical to mature their QA process a bit.

No comments: