Grayer skies for day 2. Also a bit the worse for wear from the night before. Caffeine for the soul.

DHH did the keynote – and while he didn’t say that much (information content), what he said was interesting, and well developed: legacy code makes us better programmers. We learn by seeing our past mistakes, and correcting them. Rewriting from scratch will be way less instructive than actually giving love to the old code.

Next talk i attended was about jQuery by Yehuda Katz (maintainer and co-author of jQuery in action amongst other things). Let’s say that it was a good talk for absolute beginners, but while not being an expert at all, i didn’t learn anything new.

At some point during the day, we got what was for me THE news of the conference: localization in Rails 2.2 ! Finally ! A language by a japanese person, a platform by a (former) danish citizen, so i’m happy that finally they correct the US-centric approach.

Start of the afternoon: talk about combining different apps seamlessly to improve scalability by 2 guys from Texperts. They do this by using javascript widgets pointing to different apps.

I then moved on to a talk about Rubinius by Wilson Bilkovich. Completely lost there, since he seemed to assume in-depth knowledge of how this kind of compiler-VM works, and mine is sketchy at most. They now use LLVM as a virtual machine, that they have a lot of primitives in C that are platform-specific but most of the core classes are now written in Ruby. The Ruby parser is the same as for MRI, and then *mumble Kernel mumble compiler C++*. Sounded cool in an incomprehensible kind of way.

Then the organization broke down a little bit, because 2 talks were cancelled, and so the main common-interest one, about security, was absolutely mobbed (picture geeks squatting every square cm of the room). In a fit of claustrophobia, i decided to stay out and follow the talk as well as i could through IRC (which was not too bad actually).

Some highlights: the usual cross-site scripting, but also the reading of session info in the cookie, sql injection on some params in rails pre-2.1, and then a cross-site JSON attack i’ve got to read up about because didn’t quite get it in the flow of conversation.

Sun is one of the main sponsors of this convention (together with Engine Yard),
which was already obvious by the plugging of JRuby at every possible occasion.

Nick Sieger (Dr Nic ?) did his bit for Sun, and then there was a long, long keynote by Jeremy Kemper. Jeremy Kemper is one of those guys who may be great programmers and project managers, but may not be stage material.

He talked about the performance of rails, and how he found that most of it could be reduced by looking at the browser (basically this stuff), and some in the actual garbage collection of the Ruby virtual machine (MRI).

Then we adjourned, and those of us who had been lurking on IRC to counter the slight boredom gathered to have a nice indian meal, and then some drinks at the Irish pub. A good night amongst international geeks.

In Berlin they know what summer means, and how to taper off softly to winter. The day started sunny with a hint of freshness.

So i started off towards the Railsconf in a pretty good mood, despite the early hour. I ended up subscribing to the tutorials, because i’ve been told last night (at the Bratwurst on Rails) that they were worth it.

My morning tutorial was about hacking the Rails internals by two people of the company Intridea. About showing how you can minutely twist the default rails magic to make your life easier. Pradeep Elunkamaran in the first half was fairly interesting – except that it was a bit fast, especially since (gasp) i hadn’t had my coffee yet (and hadn’t eaten since lunch the day before).

The second half was by Michael Bleigh, who’s given us subdomain-fu and uberkit (didn’t know the last one). This was more applied, and more focused towards getting a result (without knowing all the details underneath). Different point of view.

Lunch was pleasantly spent with the belgian delegation: 4 of OpenMinds, Alain Ravet, and 2 guys of belighted (and me). And two start-up founders from resp. Germany and Austria.

I started the afternoon at a presentation about deploying and monitoring rails by people from Peritor, which was woefully basic. They started off by describing all the most common server setups, and then talking about capistrano. I gave up hoping when it appeared the level wouldn’t exponentially increase.

So i skipped talks to one by Thoughtworks people about meta-programming in Ruby. Pat Farley made a good job of taking something that could have been endlessly dull (Ruby C internals) and making it bearable, grabbing our attention with humor here and there.

After dinner break, there was a panel discussion by DHH, Jeremy Kemper and Michael Koziarsky.. All being Rails core people, there was not much debate, so it became a Q&A session. About future of web apps according to them, if other test frameworks were going to creep in (no), what they thought about the other frameworks (cute but not Rails) … not many surprises there, in other words.

Rejectconf took place at a bar a couple of kilometers away, and was one of the best parts of the day. The formula was that everyone had maximum 5 minutes to talk about a related subject, however wacky

Best_of: a ‘make_spec_better’ plugin that passes all specs, whatever happens. An application by leethal, who apparently gets pretty frustrated on IRC, to be able to tell people how exactly they fit the term ‘retard’ (unable to google etc). Braid was presented as the piston for git by a young dutch guy (whose name i didn’t catch, but surely will find again).

Geoffrey Grosenbach talked about Zshell – a bit surreal to actually see the guy after listening to his disembodied voice in a few Peepcode tutorials. He then was nice (and brave) enough to lend his Macbook air to increasingly drunk presenters.