May 30, 2012
The next trip report is from Julien Laffaye:
I arrived in Ottawa on May the 8th in the evening. I went to the Royal Oak to meet with other FreeBSD developers around a beer.
On the 9th, the DevSummit started. I attended the Ports working group. We talked about the state of pkgng and its upcoming Release Candidate phase, building package sets for pkgng, and about the packages building cluster in general. We also discussed the impact of Clang being the default compiler in FreeBSD 10 for the Ports Tree and the state of the “switchable compiler for ports” patches. In the evening, we went to grab some poutine with a small group of FreeBSD ports committers. This night, I also met Tom Judge, a ports contributor and BSDCan speaker, for the fist time.
On the 10th, the second day of the DevSummit, I attended the Admin working group. We talked about FreeBSD cluster administration topics. I was especially interested by the pkgng repositories distribution architecture. For instance, do we want to build a CDN? If we build repositories every week for each supported FreeBSD version for each architecture, we can end up with terabytes of data. The project also wishes to have full access to the nodes to ensure that the mirrors are up-to-date. This will also enable us to collect some statistics.
During the break, based on what we heard, bapt@ and I went into a quest to be able to deterministically build pkgng packages. Indeed, if few packages change between builds, we can share them between the repositories and save space.
Then, I went to the Toolchain working group where we talked mostly about the state of clang in FreeBSD 10. We went back to the impact of this for the ports tree.
In the afternoon, all chairmen of the previous working groups presented a summary of what has been accomplished. This was indeed interesting because I could attend all the working groups! Then, the session turned into a wishlist: every developer talked about the feature he wants in FreeBSD 10.
The 11th and 12th were BSDCan conference days. I attended many great talks, but as I had to make choices, I could not attend them all. A very interesting talk was the one by Baptiste who presented the work we have done on pkgng so far. The fact that pkgng was well received was very motivating because I now know that we did not go into the wrong direction! The questions were interesting, and it felt good when the answer was “we already fixed this problem” and “yes, we do it that way”.
All in all, it was a very pleasant trip. I was able to meet again with FreeBSD developers, contributors and users from last year, as well as seeing new faces. It was also five interesting days for pkgng.