The great systemd bug squashing party of 2016

First, a plot of active systemd issues on github:

Active issues (open, not tagged as awaiting feedback)

We closed approx. 50 issues in the last two weeks. This is a result of some heavy lifting:

  • Lennart Poettering rewrote the unit installation logic to separate transient units into their own directory, and to support operations on a chroot in a simpler way (e.g. enabling a unit in a container from the outside). This closes a bunch of long-standing bugs where the installation logic would touch different files and symlinks than expected.
  • I later added a bunch of fixes to the installation logic and reporting, also closing a number of long-standing issues.
  • Evgeny Vereshchagin fixed various issues in the test framework and a documentation bug.
  • Susant Sahani submitted pull requests for RFEs to add configuration switches ProxyArp, MulticastQuerier and fix bugs in the networking code.
  • Martin Pitt made also tweaked to the installation logic and added entries to the hardward database. (And Martin Pitt and Daniel Mack kept our CI churning, which certainly helped avoid regressions).
  • Felipe Sateler, Nicolas Braud-Santoni, Calvin Owens, Dan Nicholson, Federico, Franck Bui, Harald Hoyer, Ismo Puustinen, Klearchos Chaloulos, Lubomir Rintel, Lukáš Nykrýn, Michal Sekletar, Michał Bartoszkiewicz, Peter Hutterer, Robert Węcławski, Taylor Smock, Thomas H. P. Andersen, Torstein Husebø, Vittorio G (VittGam), Wouter Verhelst, Frank Heckenbach, Michael Olbrich, mulkieran also provided patches in the last two weeks, most of them bug fixes, and the rest being hardware database additions, which can be considered bug fixes too :)

Are we buried by bugs yet?

To put this into wider perspective, we can look the history of issues being opened and closed. Previous graph is the difference between those two and issues "postponed" (awaiting reporter feedback):

Opened issues and closed issues, and their difference

After the initial ramp-up period last summer, the trend for new issues is approximately linear, with a hundred new issues opened each month. The trend for issues being closed is different: we seem to have longer and shorter periods of bugfixing activity, separated by periods where very few bugs are being closed. The final outcome is not too bad, with ~260/1100 ≈ 25% issues remaining open.

Pull request status

The situation for pull requests is rather different. We close almost all pull requests at a very steady rate:

Pull requests awaiting review or merging
Opened and closed pull requests, and their difference

Comments !