After a few years, I once again managed to attend the FOSDEM conference. It was a good experience, and I enjoyed most of the talks I attended.
The opening talk was preceded by an overview of the conference, the beer fest of the previous night and the FOSDEM dance, where geeks showed off their skills. The room was quite packed, with people moving in all the time during the first minutes of the talk given by Mark Surman on a Free.Open.Future.
Mark’s main points were on the successes of Open Source thus far, especially the adoption of open applications and platforms in the last year: WordPress, Firefox and the Linux powered netbooks. He stressed that the rise of multiple browsers has helped getting rid of websites that are only willing to serve pages to the Internet Exploder. Of course, code also needs content to make things interesting for people. The main example Mark talked briefly about was Flickr and the mass of Creative Commons licensed images.
Yet the road ahead seems to be one where Open Source is heading into a very closed environment: the mobile world. It seems that the mobile environment will need a different conceptual map from what we have been using in the non-mobile computing world. Pressure points to get past the closed ecosystem, where mobile phone operators dictate law, are still largely unknown. Mark claimed we need better and bigger maps to gain an understanding of how the mobile world moves, who the payers are and how we can break through misconceptions. Several things probably ought to be on the map: strong values, freedom beyond just code; free software that empowers people (even those who do not agree with the free and open idea); users as hackers.
After the talk, I went out to peek at the booths, after which I returned to listen to the remainder of the Debian Talk. Seems like a release is due by February 14th. We can only hope.
The afternoon started with a few talks on the state of the OpenJDK, IcedTea, and such. The Free Java DevRoom was packed. I doubt were conformed to the fire department regulations.
I had expected a lot off information on the state of the VM itself, but all I got was a short list of stuff the VM would incorporate: a new GC, compressed 64-bit pointers, and several HotSpot runtime compiler improvements. The OpenJDK 7 will be lacking closures, first-class properties, operator overloading and bigdecimal syntax. Joe Darcy proceeded with a short overview of OpenJDK 6, for which an open Mercurial repository outside any SUN firewalled domain was created. Mark Reinhold took up the presenters place talking about JigSaw, and told us most of the stuff that was put in this blog post. I mainly retain that they want to get rid of unscaling jar files and fully modularize the whole ordeal.
The last talk I attended on Saturday was a short intro to Scala where Martin Odersly tried to convince the crowd that the Scala approach was the way forward. I pretty much liked the fact that it adopts a functional approach to programming, in some given sense of functional. I list the main points that drew my interest: strong typing, type inference, little boilerplate are the basis. Scala has higher-order functions, algebraic types, parametric polymorphism and subtyping and inheritance, dynamic configurations, classes as partial abstractions. Yet it does have side-effects (not encapsulated), yet it uses functions as first order objects (currying, partial application, …) Functions in Scala are also objects, embodied by traits. This means that functions can be subclassed or specialised, e.g., for arrays, collections, etc.
For the moment, however, the only advantage Scala seems to have over e.g., Haskell is that it can fully reuse the entire Java codebase that floats out there. Still, I think Haskell is still the better programming language. And a number of geeks certainly seemed to agree as the RWH book sold out ofter a single day. Of course, O’Reilly might not have carried a whole lot of them to begin with, but I thought it neat nonetheless.
On Sunday I arrived early, not wanting to miss Dries’ talk on Drupal 7.
Currently Drupal 7 is still in a flux, with a code freeze months away 
The biggest issue that is being tackled is usability, which is critical to adoption. A lot of small things are already in: password strength checker, installation profiles where much used features are set up automagically. As Dries mentioned a while back, Mark Boulton will do some work on the usability front. Cool is that files have become first class citizens, with hook readily available to plug into and get notified when files are accessed, changed, etc.
The major point, to me, was the move towards more semantics in the output of the Drupal engine. This should allow machines to more correctly interpret what each piece of data represents and how the user will perceive it. Another major issue is getting the CCK to the Drupal core. Dries mentioned that this move has driven a lot of innovations. Fields are now in core, which enables a lot of other things. Thus, a node can have Fields, which can be (almost) anything. Basically, the CCK is now reduced to the UI in contrib.
By Sunday noon, the stack of Using Drupal books had also dropped to a single item.
I noticed there was some love/hate relationship with Gnome:
In the afternoon, I attended the Ext4 talk by Theodore Tso, which seemed to indicate we’re moving back to an old scheme where contiguous regions are stored in the inode rather than addressing each block separately. I wonder how well this scheme will hold under a full disk, or a disk where a lot of stuff gets written and deleted. Apparently there will be a defragmenting tool running in the background. I am still partial to the whole idea.
I remained in the Janson room to listen to Kris Kenneway, a FreeBSD hacker, talk bout peformance analysis in FreeBSD. It was a good talk, although it started a tad slow.
The last talk of Sunday on the Google SoC was also one of the best, except for the typically American style where the crowd is supposed to go crazy and, as Leslie Hawthron put it ‘feel the love’. I thought not But, the presentation was nifty and she managed to convey her message well, I think:
And so it was. But in the face of overwhelming odds, people can thrive. So did the group at Google that runs the SoC.
All in all, a fun experience. I even learned a few things, and if my schedule allows it, I’ll be back next year.
Tags: brussels, conference, fosdem