Joachim Breitner

Reproducible Builds World Summit

Published 2015-12-04 in sections English, Debian.

This week, I was attending the first Reproducible Builds World Summit in Athens. A while ago, I have fixed some reproducibility bugs in the Haskell compiler GHC (parts of #4012), and that got me a ticket to this fully sponsored and nicely organised event.

This was the first event I attended that had a dedicated and professional moderator, Gunner from Aspiration, who came up with a suitable on-the-fly-schedule, ensured the time was structured by shorter plenary rounds with all ~40 attendees as well as longer working sessions in groups of usually at most 10 people, motivated people to run (“facilitate”) these working groups as well as ensured that any results were noted and collected. This definitely makes a difference to the otherwise unstructured “let’s all sit in a big circle and endlessly discuss things“ or “let’s all just hack on what we want to hack on right now” structure of less professionally organised meetings.

But after 1½ days of talking, a certain desire to hack and get work done could be felt throughout the group, so most of us were happy to be allowed to make use of the laptop on Wednesday afternoon. At least as far as possible – the wireless was not expecting this crowd, it seems.

Besides the few patches above, my involvement in the project is rather rudimentary, so I tried to contribute as far as possible during the event itself. This was not always easy, as many of the working sessions were either self-reflecting or forward-facing: What problems do we see? What should we get done next? How do we talk to users/upstream/press/doners? Where possible, I tried to at least be helpful in the discussion.

During the hacking sessions, I found it most useful to contribute to diffoscope, a program to recursively and deeply show the difference between files of any format, so there are now patches to implement a multi-file HTML output, to more efficiently include large diffs, and – more importantly – I assisted Lunar in making diffoscope multi-threaded. In Python this is not a very great experience; I guess I am spoiled by Haskell (as it is often the case).

I enjoyed the group; a bit like DebConf but with many new faced from similarly-minded, but different projects. It was, however, again a Free Software event with an embarrassing low number of female participants. I liked discussing a proof of Gödel’s first incompleteness theorem during a skill exchange session. Unfortunately, I had to leave very timely after the event on Thursday evening, in order to attend the Haskell in Leipzig workshop.


Have something to say? You can post a comment by sending an e-Mail to me at <>, and I will include it here.