DebCamp for me extended an extra day to include the DebConf18 open day, but fortunately no further; today is the first day of DebConf18 proper, and I am happy to have DebCamp work sufficiently wrapped up that I can go to talks and discuss past and future work … rather than actually doing that work.

Here’s what I did at DebCamp:

• Significant progress on both difficult and easy bugs against Debian Policy, and some admin issues. I haven’t managed to make a release but I think enough will fall into place between now and the end of DebConf that I can do that before flying home.

  Hideki Yamane is working on adding the infrastructure to translate Policy, which is very exciting. I don’t think we’ll get that all finished this week – I have to learn how to co-ordinate translators – but if we can get that infrastructure merged and released that is a big blocker out of the way.

• Developed and released dgit versions 6.0 and 6.1 with Ian Jackson. The main changes: --build-products-dir works properly now and is configurable in your ~/.gitconfig, and dgit gained built-in support for pbuilder.

  There are various internal improvements to make these features work quite elegantly. Some of the work was reworking parts of a patch series I posted a few weeks ago, so it should be easier to rework the remainder of that series now.

  We realised yesterday evening that while the --build-products-dir work was intended mainly for the benefit of pbuilder users, it is actually quite a boon for sbuild users too, because sbuild itself does not support dumping .orig, .dsc, .debian.tar.gz etc. files into a build products directory. But now it does, if you run it via dgit.

  I have never been bothered by these build products in .., but I am considering starting to use the new feature and throwing away my crude perl code that cleans up ~/src on my workstations.

• Finally found a place in the archive for some mail processing scripts in my ~/bin: a new package called mailscripts (after some discussion about where else they might be included). And finally wrote, and included in this package, a facility to pull a Debian bug into your notmuch database. No more opening up the web browser to read bugs because you were CCed on only a few of the messages in the bug.

• Provided people with feedback: on a proposal for a procedure for salvaging packages; on Ian Jackson’s script and slides for our talk about git-debrebase.

• Small pieces of work on team-maintained packages. Filing RFAs and removing myself from Uploaders where I’m not active. Some sponsorship. Some NMUs to help a transition.

• Some ordinary package maintainance.

• Some unfortunate yak shaving rabbit holes. Including broken /etc/mailname on my laptop. But that’s better than David Bremner, whose entire setup for sending e-mail broke in the middle of the week.

There has been a very positive and almost entirely non-hierarchical division of labour all week: the event organisers, who are all just Debian volunteers, have been dealing with quite a number of challenging issues, but these have not filtered through to bother those of us trying to power through work to improve Debian.

Further, the organising team has expanded to include people who had earlier in the week seemed a bit lost and didn’t seem to have something to do: a beginner data scientist from Taiwan who had showed up at the conference seemingly not knowing too much about Debian, and a group of Eastern European interns who have been contributing to Debian or related projects through schemes like Google Summer of Code and Outreachy. They all donned red staff T-shirts and got on with the job.

It’s not that these organisers are working for those of us writing code and discussing policy. By enabling each other’s work we are all working for the good of Debian, and thereby the much wider good of anyone who directly or indirectly interacts with computers. I have been seriously impressed by how effective and non-hierarchical this division of labour has been this week. But now I’m happy to have some more time to actually talk to the organisers, instead of running off to try to fix a test suite failure.