Over the past few months there have been email discussions and complaints on IRC about the Red Hat Bugzilla instance and its shortcomings. It is often suggested that these shortcomings could easily be overcome by creating a separate bugzilla instance–separate in the sense that a separate instance would be divorced from Red Hat, only contain bugs related to the Fedora Project and managed by Fedora without having to go inside Red Hat.
It has been proposed that a stand alone bugzilla instance would:
- increase performance
- enable better search
- make it easier to file bugs
- be easier to customize
I am not completely against a separate bugzilla instance for Fedora, however each time I see these arguments raised I wonder we if we are focusing on the right problems at the right time. Have we clearly enumerated exactly what problems we are trying to solve and are they the right ones?
Several years ago in between changing careers I spent several months in Europe traveling and taking time out to make better sense of my life. During this time I met a wise person in the Swiss Alps who reminded me that changing your location–as amazing as it might be–rarely has the affect of permanently changing whatever it is you want to be different. Because as he put it, “you bring yourself with you.”
If we change to a new instance of bugzilla, we still bring our old bug handling practices with us, which no matter how new or shiny the tool, will remain unchanged unless we make a concerted effort to do so.
By this I am referring to how Fedora triages and resolves bugs and the processes–formal or informal–to achieve this. Presently there are approximately 9,000 outstanding bugs in a status of NEW, ASSIGNED, or MODIFIED for Fedora versions still supported–an increase of 2,000 bugs using the same criteria compared to March 2007. The number of bugs for supported releases increases to approximately 10,000 querying all statuses except CLOSED.
Simply making it easier to add more bugs and search the bugs we have will not magically help resolve theses 10,000 bugs and make Fedora a better distro.
I was asking Dave Jones his thoughts on this problem at OSCON back in July 2007. From the perspective of the kernel, one thought he had was to have kernel bug days and get community help with triage. He recently took things one step further by creating a very clear and helpful wiki page about how to triage kernel bugs. I don’t know that much about the kernel, but it was encouraging to see that there were some basic ways to start helping.
Triage is just one aspect of managing bugs. We also need to consider some type of auto-close mechanism or process for bugs that are stale or will never be fixed. Some of the performance problems and other current deficiencies of Red Hat’s bugzilla should go away once it is upgraded to Bugzilla 3.0. Maybe in the end a separate Fedora bugzilla instance is the way to go, but I wonder if our initial time and energy would be better spent on deeper problems first.