I’ve been to FOSDEM this year (again) and as always it’s been an interesting experience. You meet all kinds of people, you can create new contacts and realize different projects can share some code or at least approaches. This is just going to be a rundown of talks that I’ve seen (or most of them anyway).
(R)evolution of Java Packaging in GNU/Linux
I’ll start with talk Mikolaj Izdebski and myself gave in Java devroom titled (R)evolution of Java Packaging in GNU/Linux, since that resulted in some possible interesting cooperation. The talk went reasonably well without any hiccups. We could have used more time for discussion and generic talk about Java buildsytems, but oh well.
After the talk Charles Oliver Nutter (headius @ JRuby/JVM projects) mentioned Sonatype is already running conversion of all jar files to rubygems on their infrastructure. His idea (which seems pretty neat) was to make this for RPMs as well. He got us in touch with Jason van Zyl and it seems if we create conversion tool we might get this to run on their infrastructure. This actually ties in nicely with some work we are doing to improve life for Java developers on Fedora because they’d be able to install unlimited number of different Maven artifacts that could tie in directly to Yum/RPM and companies could limit access to Maven Central repository and just allow these RPMs.
The state of OpenJDK & OpenJDK7u & OpenJDK Q&A
These were talks by Oracle guys giving updates where JVM is going, and how does the schedule look. Nothing too interesting or worrying here, but it seems that Oracle is really trying to put more people working on JVM, plus there’s community participation as well. Mark Reinhold’s prediction was that OpenJDK board will stay the same after reelections in next few months.
Interestingly Q&A at the end of Saturday was waaay more calm than previous years. Maybe that has to do something with community around OpenJDK maturing?
Porting OpenJDK to AArch64
One of the more technical talks about intricacies of simulating hardware that (almost) doesn’t exist. Our Andrews (Andrew Haley and Andrew Dinn) had some hard drive issues but they handled it pretty gracefully. The trick with jumping between AArch64 and x86 code outside of simulator was quite interesting. Also interesting were notes on register usage by JVM and how number of registers helps or hinders implementations.
Community Management in Meat Space
Talk by our very own Leslie Hawthorn and Lydia Pintscher. Basically it was a quick tutorial for dealing with decision making and problems within communities. A sort of small prelude to Leslie’s keynote. It just had more duct tape :-0
This talk was done by one of more controversial people (one of co-developers of eudev which got a lot of flaming apparently). Francisco Riera used a volunteer to show various features of hardened kernels (not necessarily SELinux style) in Gentoo, how and why they operate in certain ways. Each change was accompanied with quite nice examples how in non-hardened kernels certain things could be misused. I missed last few minutes due to other talk.
Jeremy Allison talked about history and splitting of of Samba4 from Samba3 codebase. Apparently there was almost a fork due to “old” network filesystem guys and new “let’s create an AD” guys. Code still has samba3 & samba4 source directories but they are merging/cleaning them up. They have a lot of code bundled/developed in Samba4 to ease integration and configuration. Kerberos, LDAP..the list goes on. Makes me wonder if all of this was *really* necessary. I guess that’s why Jeremy added this bit to one of the slides:
We stopped checking for monsters under our beds when we realized they were inside us.
systemd, two years later
Surprisingly no flamewars. During the talk I (perhaps) understood why Lennart is facing so many of them though. I believe he doesn’t clearly state that “Yes, we reimplemented part of X in systemd so that it’s more reliable. BUT! You can still install that old thing, we won’t break it and you can still configure it”. Prime example for me being acpid, which Lennart replaced just to get power button working. That’s probably fine for 99% of use cases, but on a lot of notebooks acpid is primary way to handle interesting buttons that X knows nothing about (and perhaps we need them to work outside of X)
FreedomBox is a project by Eben Moglen and Bdale Garbee, now mostly a software solution running on top of Debian systems to make private and secure communication without dependence on governments. And one important feature: it has to be usable by common people. There were already 2 keynotes about it in previous years. Now it’s slowly coming to 1.0. There was an interesting point about replacing CA infrastructure in webservers with p2p/gnupg trust principles and development of apache module that would be able to authenticate these.
LibreOffice: cleaning and refactoring a giant code-base
I really like where LO is going. They’ve been mostly doing huge cleanups, getting rid of old cruft, German comments etc. Java dependency is going away. They have a healthy community, ood code review/unit test processes. All the right pieces. I can’t wait for when they really start adding new things.
Has GNOME community turned crazy?
Talk about various controversial features developed by Gnome is past few years. Mostly Gnome Shell. Vincent Untz made a good point: Gnome 2.0 was nothing like Gnome 2.3x. However when he asked audience if they thought Gnome 3.0 was ready for release even people who like Gnome shell said no. There were a lot of Gnome people in the audience who were co-answering questions.
The Keeper of Secrets
Our Leslie Hawthorn had a closing keynote, which in my mind was a continuation of her previous talk. It dealt with handling community participants when they have problems they would like to keep secret while still allowing the community to handle their absence. Most valuable part for me was probably resources & references to other literature. Because let’s face it…you can’t really give a silver bullet generic answer, because each situation is different. I guess it could have been articulated more clearly, because some people in the audience never realized this in my opinion.
Now I can start looking forward to DevConf already.