Monthly Archives: December 2018

Correlations Between Firefox Bug Severity and Priority

I was curious whether there were any correlations between Firefox bugs’ Priority and Severity values. My hypothesis was that:

  1. The Severity field would be rarely set to a non-default value. (Mozilla typically uses separate status flags to track blocking bugs and low Priority (P3-P5) for non-urgent issues.)
  2. Priority values would be correlated with Severity. (If a bug is severe, it will probably have a high priority.)

So are there any correlations between Severity and Priority values?

  1. The Severity field was rarely set to a non-default value: about 90% of triaged bugs had the default Severity (“Normal”), regardless of Priority. So Severity was not correlated with Priority.
  2. However, Priority was correlated with Severity: about 85% of triaged bugs with Severity “Blocker” had Priority P1. About 30-50% of bugs with other Severity values had Priority P3.

There are a lot of Firefox bugs in Bugzilla. To narrow the scope of my analysis, I selected bugs from the Firefox desktop product (Bugzilla Product “Firefox” or “Core”) that were filed and triaged within the last two years.

I ignored bugs filed more than two years ago because Mozilla’s use of the Priority field has changed over time. The last two years roughly cover the current era of bug triage and prioritization practices where Priority values have the following meaning:

  • P1 = Fix in the current release or iteration
  • P2 = Fix in the next release or iteration
  • P3 = Backlog
  • P4 = There is no P4. (This Priority is not supposed to be used, though it still is.)
  • P5 = “Patches accepted”

I ignore bugs with Priority “–” (the default and untriaged) and P4 because these values are not set by triagers. P1 bugs may be overrepresented because, in theory, P2 bugs are not supposed to be fixed in the current release and should be elevated to P1 to be scheduled for fixing in the current release. I also ignored bugs that were resolved as Invalid, Duplicate, or Wontfix because they are unlikely to have accurate Priority or Severity values.

Distribution of Priority values:

  • 20.36% P1
  • 18.76% P2
  • 47.56% P3
  • 13.32% P5

Distribution of Severity values:

  • 0.15% Blocker
  • 5.53% Critical
  • 0.87% Major
  • 89.39% Normal (the default)
  • 0.87% Minor
  • 0.31% Trivial
  • 2.88% Enhancement

In conclusion, I propose that the Firefox Bugzilla not even show the Severity field. It’s not used to track which bugs block a given release or which bugs should be fixed first. For my raw bug data, see this spreadsheet.