I had the pleasure of attending and speaking at DevConf.cz this year. It was also my first visit to the Czech Republic, which was fun for me since my grandfather's family originates from Radnice.
Friday was the first day of the conference. We got up bright and early (well, maybe not bright…) and headed over to the venue. I spent a fair amount of time on Friday attending talks.
I started with the keynote, presented by a variety of speakers representing a wide range of Red Hat's products. The keynote told a narrative of going from unboxed, racked servers to deploying code live from Eclipse to production on those servers (and all the steps in between).
Next I attended "Generational Core - The Future of Fedora?" by Petr Sabata. Petr presented about Fedora's modular future and how Factory 2.0 fits into the picture.
I then went to see Adam Šamalík and Courtney Pacheco give us another modularity talk. This talk presented more detail about the build system and design of Fedora modules.
I spent the afternoon eating a delicious hamburger provided by the conference. Well, not really the whole afternoon. I snuck in a little bit of hallway track time and worked a bit with Pierre-Yves Chibon on the talk we would give on Saturday.
I started Saturday by attending the keynote talk about Fedora, Red Hat Enterprise Linux, CentOS, and how they are all better together. It was a wonderful narrative about how each of the projects/product contribute to one another's success. Well done!
I spent Saturday morning putting the finishing touches on my two talks that I'd give that day. The plans for how we would mirror container images had been in flux, and the slides I had prepared weren't quite accurate anymore so I had to put some last minute corrections into them.
I then gave both of my talks, back to back! It was a rush. I felt really good about both talks. There seemed to be a lot of interest in the Bodhi talk. Pierre demonstrated some neat ideas for integrating Pagure and Bodhi, and we have a pretty lively Q&A session. I had to rush over to my next talk, which was in a different building. This one had a less lively Q&A session, but I think it went well too. You can more about my talks here.
After my talks were done I had a rush of relief as I no longer had to worry if I'd done enough preparation for two talks (note to self: perhaps only pitch one talk next year…)
I spent the afternoon hacking on ejabberd with the legendary Peter Lemenkov. He had been working on some patches to add Kerberos support to ejabberd and wanted to try them out. We built some boxes in Fedora's OpenStack instance and got started with patching. We quickly realized we were in over our heads as neither of us had much server-side Kerberos knowledge, but after a few SMS messages we were able to enlist the help of Patrick Uiterwijk, a wise security sage. He created us a keytab in Fedora's staging environment and got us moving again. We never got it working on Saturday, but we did make a lot of progress.
There was a neat survey/test being done in one of the conference
hallways about the
openssl command line tool. They had you perform
some tasks and ask you about the tool afterwards. If you were able to
perform the tasks to their satisfaction, you were awarded with a hat. I
got a nice Red Hat toboggan!
Saturday night was the conference party, which was a lot of fun. I got to spend time with a lot of my coworkers from all over the world in a more social setting.
Sunday morning's keynote was a very entertaining argument between Dan Walsh and Steven Pousty about the future world of containers. It was my personal favorite talk of the conference and I highly recommend watching it if you haven't seen it.
I next went to see Till Mass talk about Certificate Transparency. It was a very well delivered talk with very nice slides. There were some unfortunate projector problems, but Till did a great job of navigating the technical difficulties. Certificate Transparency was news to me, but it creates a public database of all issued certificates (from participating Certificate Authorities). This would prevent a compromised authority from issuing a certificate for a CN that has been issued by another authority. It addresses what is currently a significant weakness in our public key infrastructure.
After a break I watched Dennis Gilmore talk about "Moving everyone to rawhide". I've been running Rawhide on a few systems here and there, so I was interested in hearing his perspective. He shared a lot of changes that are coming to Fedora in the future, such as Fedora 27 not having a beta, and Fedora 28 might be the last versioned Fedora release (whoah!).
I then had another hacking session with Peter Lemenkov on ejabberd. It turned out that Patrick Uiterwijk was able to identify some problems on the ejabberd server side and they got it working together. I then worked with Peter to clean up our implementation a bit and do additional testing. It seems to work well. We are considering deploying an ejabberd to Fedora's Cloud (i.e., it would be an unsupported community run service), but there are still a few things we'd need to figure out.
I saw Christian Schaller give a talk about Fedora Workstation. He talked about some of the changes that are coming in the future, and the challenges the workstation team faces. There was a lively Q&A session afterwards.
After that I did a little bit more hallway tracking, and then I went to see the concluding session which was a fun quiz with prizes.
The conference was a wonderful experience, and I look forward to sending a talk for next year's DevConf (though maybe I'll just do one this time ☺)