Tuesday, July 7, 2015

BSDCan 2015 Trip Report: Warren Block

BSDCan 2015

By Warren Block: wblock@FreeBSD.org

BSDCan, the large BSD conference in Ottawa, Ontario, was held in June this year.  Each year, it becomes a bigger event.  As in previous years, Andrew Ross was there with the FOSSLC Project, recording the presentations with excellent video and audio quality.  As of this writing, some but not all of the recordings have made it to YouTube, and links are included for the ones that are known.

Developer Summit
A FreeBSD developer summit is usually held in the two days prior to the conference itself.  The summit gives developers a chance to plan projects and present ideas to other developers in person.

I was lucky enough to give the documentation working group's presentation.  There is always concern about whether enough people will be interested in any given topic to make the meeting worthwhile.  That turned out not only not to be a problem, but we actually had extra people show up.  Better yet, they were all enthusiastic and interested.

We talked about revamping our translation system and tuning the documentation toolchain, possibly asking for help from the Foundation to do that.

Another major point was encouraging people to contribute to the documentation, and lowering the barriers to make such contributions easier.  Live demonstrations are the computer equivalent of "Hold my beer and watch this!", yet I managed to show how Annotator (http://annotatorjs.org/) could be used to let people review and comment on our documents without having to know any of the complex markup languages.  Help is still needed with porting a Storage component for this, please contact me if you are interested.

The FreeBSD wiki is underused, and we talked about ways to allow more people to contribute there without lowering our standards of quality.  I actually forgot to mention my idea about that, where new wiki contributions would automatically generate reviews at https://reviews.freebsd.org/.  Either a special group of wiki reviewers or any committer could approve them, spreading the load.

Another big issue was the FreeBSD.org website.  Almost everyone agrees it needs to be updated.  Where they disagree is the is the issue of how. The Foundation has updated their site recently, and might offer help with ours.

We had a total of three hours for the documentation working group, used all of it, and could have used more.

Doc Lounge
On nights after the developer summit and BSDCan, the FreeBSD documentation group holds an informal session where people can gather and learn about or work on documentation.  We usually have a few short presentations and try to provide one-on-one time for attendees to talk with documentation committers.  The trick is trying to work with all the people who show up.  The doc committers are usually outnumbered, but people are always incredibly patient.

Presentations
Only a few of the talks will be highlighted here, but go watch all of them online.  Really.

The keynote presentation was Stephen Bourne's excellent talk on sh.  It gave an inside view of software design within the limitations of available hardware.  Along with that were some insights on things that might have been done differently in hindsight, or not at all.

The opening session ran somewhat long, and there was some confusion on schedules for the rest of the day.  I gave a second presentation, called "Thinking About Installers: Discord and Happiness" about how a different view of installers could make the job of installing FreeBSD much easier, regardless of the amount of customization needed on the target system.  Despite the schedule confusion and this talk not even being listed by name on the program, it was surprisingly well-attended. Some of that was certainly due to Kylie Lang's preceding talk about FreeBSD on Hyper-V and Azure.  Still, the attendees were interested, asked good questions, and offered suggestions.

In the afternoon, Ryan Lortie of the Gnome project talked about jhbuild. Gnome has recently included FreeBSD as a supported platform, and this is their method of continuous testing and building.  Initially met with not much interest, the interaction and cooperation has grown and benefitted both projects.

Later in the afternoon, Joseph Mingrone showed the use of FreeBSD clusters for genomic analysis. The application of clusters to this high-end scientific project was interesting, and combining it with down-to-earth sysadmin problems made for a unique presentation.

On the second day of talks, Kris Moore talked about shifting the PC-BSD management utilities to web applications.  That was interesting, but then it got even more interesting when he described background updates, which make it possible to upgrade a machine transparently while in use. This builds on the concept of ZFS boot environments, and has serious advantages for end-user desktop systems.

Baptiste Daroussin gave a fascinating talk about packaging the base system.  This will be a big step forward in being able to install and customize FreeBSD systems quickly and easily. People have been concerned about how this change might affect users, but the audience was very receptive and the talk anticipated and answered those concerns. One of my former mentors, FreeBSD Release Engineer Glen Barber, even spoke one short sentence into a microphone, probably breaking his previous record.

Later that day, Sevan Janiyan (who we met last year at the doc lounge) had a very interesting talk about building pkgsrc applications in unusual environments.  pkgsrc is NetBSD's equivalent to ports, and he took that tree and attempted to just get it to build on numerous Unix variants and architectures.  Even just trying to build the tree without succeeding caught bugs, often serious ones in the host systems.  And that was from just trying to build the tree, not get the applications to run afterwards.  Several of the talks this year pointed out how revealing testing could be, even trivial testing.  Testing is good.  We need more of it.

During this talk, we had our second schedule confusion.  A fire alarm went off, and we had to evacuate the build for a short break.  It turned out that someone had overheated something in a microwave on one of the higher floors (my guess is microwave popcorn, the bane of office workers everywhere).  After the fire department left, we resumed the talks.

The closing talks and charity auction were somewhat rushed, but still great. The closing party at the nearby Lowertown Brewery was excellent. These informal events are great for meeting people that were only known by their email addresses previously.

The FreeBSD Foundation
The Foundation always has a table set up during BSDCan.  They are too shy and polite to collar people walking by and ask them to help support FreeBSD with a donation.  Unless you go to one of their talks, you might not even know that individual contributors are needed to show the auditors that the Foundation is community-supported, and even small donations increase the contributor count.  If you see them at a conference, stop and talk to them.  Make a small donation, or a large one if you prefer.  Many FreeBSD improvements have been possible solely because of the Foundation's support.  They continue to grow by adding good people.  This year, they announced that Benedict Reuschling, my other mentor, has joined the board.

Conclusion
BSDCan is not what most people expect.  It is not a boring computer conference.  Sure, there are presentations and talks and the standard conference stuff.  But this is a place where BSD nerds are the norm, not the exception; where the other people there speak your language, share and understand your problems, and know that you understand theirs.  It is a feeling of family, a chance to share and solve problems, and gather inspiration for the rest of the year.

Thanks
My thanks to the FreeBSD Foundation for making it possible to attend BSDCan, to Dan Langille for creating and running BSDCan, and to Dru Lavigne for always being there.

No comments:

Post a Comment