Myself and the Githubbers from Fabio Akita on Vimeo.

QCon is finally over. It was a great venue, with great people. Thanks to Floyd, InfoQ and all the organizers and sponsors. I think it accomplished its goal of discussing the new trends in technology. It is clear that functional programming, non-relational databases and the Agile philosophy are the way to go.

Yesterday I had a great time with the awesome guys from Github. Chris Wanstrath, Scott Chacon and Tom Preston, together with RailsEnvy’s Jason Seifer. We discussed languages, Git, and several geeky things :-) Thank you guys, you’re kicking ass, keep up the great job.

Ezra Zygmuntowicz was also really nice to invite me to visit Engine Yard’s headquarters. It was really interesting to finally get to know them. He showed me the new service they’re building on top of Amazon Web Services, a way to easily create your own highly scalable Rails/Merb infrastructure. They are building a very slick Dashboard so you can control all of your slices, configurations, and environments. It is supposed to be released at the end of this year. I know people will love it.

Now I am preparing to check out and head to the airport, back to Brazil after a busy but very rewarding week. See you all at the next conference.

Cheers!

So, in the end we did a 2-part interview, almost 1 hour each. You will agree that this is the geekiest interview ever. It was actually more of a lecture, with Ola Bini explaining every conceivable programming technique and paradigm in the book. Seriously.

We went through Lisp, Erlang, F#, Haskell, Java, Self, ML, Ruby, Python, Javascript, Io and much more. It was a very intense conversation so prepare yourself for an overdose of language geekiness discussion.

I had 2 goals in mind with this. First, to introduce many programming concepts before talking about Ola’s new language implemented on top of the JVM: Ioke, an Io-inspired language, prototype-based, highly dynamic, based on Io, Lisp, and Ruby. This language is way cool, you should experiment with it.

The second goal was to show people that there is this whole world out there, outside of plain Java or C#. And another thing was to not show a white-bearded senior developer like Kent Beck or Tim Bray :-) No offense, but it is accidentally convenient for me that Ola is so young (early 20’s), because now young CS students don’t have the ‘age’ excuse for not knowing all of these concepts already.

So, it was a very productive interview. Download the first audio file from here and Part 2 from here.

Today was a pretty busy day of interviews. This afternoon I was able to first interview John Straw. He is responsible for what he calls The Big Rewrite project. The project about replacing 150k LOC from Java, with no tests, to around 13k LOC of Ruby on Rails, with almost 100% test coverage, and without reducing the scope. The original project was developed over 22 months and the rewrite took place in 4 months of development, with 4 developers (though they had 4 months of preparation and planning, but still …).

In this interview he talks about the motivations, how it was for the team to move from Java to Ruby, how they chose Rails, what’s the size of their infrastructure. It is a great case study for any company using Java to be reassured that changing to Ruby will only bring benefits.

After that I finally interviewed Matt Aimonetti. He is the main Merb Evangelist. He has a training and consulting firm in San Diego, he was also responsible for MerbCamp, the first Merb event around. And he is also one of the main contributors to Merb.

He was kind enough to spend a long time showing me the nuts and bolts of Merb. I was not aware of its current state and I have to tell you that it is pretty compelling. Very well thought out, it has everything you need to start developing web applications with almost the same ease of use and convenience of Ruby on Rails.

Among the best things I saw in Merb: it can be pretty close to Rails, so you will feel right at home. It has “Slices”, which is a feature that I expected Rails to have for a long time – it works almost the same way as Engines, but it is built-in and feels much better. It has a neat feature of a “master process”, so you can instruct it to load N worker processes (such as a mongrel cluster) and it will monitor those workers, so if one goes down, the master will respawn it automatically, which is pretty convenient. And finally, its modularity is top-notch. It feels weird at first having lots of gems around, but it makes sense very fast.

And according to Matt, Merb is way faster than Rails – at least in a “Hello World” benchmark :-) All in all, I highly recommend it, especially if you’re already an advanced Ruby developer that wants more (or less) than Rails can offer out of the box right now.

Download John’s audio file from here and Matt’s file from here.

Nick, Matt, Chris and Jan

After that I was able to interview Francesco Cesarini. He gave us an introductory tutorial on Erlang earlier this week. He’s been working with Erlang for the last 15 years and he is also writing a new book for O’Reilly, called Erlang Programming. This is a very insightful conversation on functional programming, scalability, concurrency and why all these subjects matter today and how Erlang fits in.

It was interesting because both Kent Beck and Tim Bray were talking about future trends in their keynotes and both mentioned CouchDB and Erlang as great stuff. You want to stay ahead of the curve? Learn Erlang.

The audio files will show up in the Ruby on Rails Podcast Brasil feed soon enough, but before that happens, you can download the audio files directly from here. Click here for Nick and here for Francesco.

Another great day at QCon SF, and this morning I had the pleasure of interviewing both Jan Lehnardt and Chris Anderson, both committers for the extraordinary CouchDB project. And there is another nice twist to this as Chris is also the creator of the CouchRest project, the Ruby library to consume CouchDB resources that Geoffrey Grosenbach presents in his CouchDB screencast.

I’ve been saying that Functional Programming and Non-Relational Databases will be the way to go into the multi-core, multi-server parallel world. We are seeing this movement already. Sun is investing in different languages, including Clojure. Microsoft has been developing F# and will add functional aspects to C# 4.0. In the Cloud space we see Amazon with SimpleDB, Google with BigTable and Microsoft Azure with SQL Data Services: none of them are relational.

In the Ruby community we’ve been dabbling around Erlang for a while now, I’ve seen people trying out CouchDB with Ruby projects, even here in Brazil. So I think Rails/Merb + CouchRest will be a really nice way to have highly scalable applications almost “out of the box”.

We’ve been good at scaling the Web tier. We understand HTTP, we know load balancing techniques, we understand shared-nothing architectures. But there is always the last mile: the database tier. SQL Server implementations such as MySQL scale very poorly. Bi-directional replication is a pain to do, queries are not easily parallelizable. At some point you will have to leave the relational theory behind and start denormalizing like crazy. And at some other point, you might even need to shard your database. All this requires you to change your application code and everything is just one big and nasty nightmare.

Database scalability does not come for free, and one solution may be to leave RDBMS completely. I am not advocating dropping SQL for everything and going CouchDB, but instead that some Use Cases may be better served with Document-Oriented Databases instead.

Jan and Chris were really nice to give me the opportunity to interview them on the ins and outs of CouchDB. Bottom line: it’s ready for prime time right now. New features are coming, but you can take advantage of it today. Again, the audio file will be available in the feed for the Ruby on Rails Brasil Podcast (in English), but you can download it directly from here.

There’s been a lot of buzz around Rails lately. In particular, DHH published a series of Rails Myths articles. The #2 Mythbuster, for instance, brought Zed Shaw back for some more though it was rectified already. But Mythbuster #4 got some replies from Yehuda Katz, the current maintainer of Merb.

The main issue is around Modularity. DHH’s point of view is that Rails is modular enough and you can let some components loose. But Yehuda’s argument is that it requires you to patch Rails in order to release it from some of its components, whereas Merb was built around the concept of modularity from the beginning. Some people got worried that another Cold War was starting but Yehuda says that this is not the case and that open discussions like these are actually good, instead of having behind the scenes nitpicks.

While in San Francisco for QCon, I was able to interview Yehuda on these matters. By the way, thanks a lot to him and Leah for the nice dinner. I think this is a pretty comprehensive overview of Merb, DataMapper, and the current issues around Rails modularity.

On the other hand we had some sad news yesterday as well: Engine Yard was forced to lay off lots of Rubinius developers. Evan explained the reasons in his blog, and Yehuda states again that Engine Yard is still committed to Rubinius. Besides that, they also announced yesterday that EY will have a new line of services around Amazon Web Services, providing tuned appliances for optimal Rails deployments in the cloud.

I have all the details in this special episode of the Ruby on Rails Podcast Brasil (in English). It is not on the official website yet, but expect it to show up in the feeds by tomorrow. Meanwhile you can download it directly from this link

I’ll have more insights from QCon later this week, stay tuned.