Saturday 5 April 2008

code == data, jRuby rocks and ohh BTW i killed a villager

I'm just back form my first conference, Scotland on Rails (or I was when I started writing this blog entry... it's been a hectic week, believe me). It was great, that is to say I really enjoyed it. There was a diverse range of talks, from an experience report by first time speakers Richie McMahon & Maria Gutierrez to the innovative Edgecase dialogue from Jim Weirich and Joe O'Brien. So here is a my highlight reel.


Beautiful Code (Keynote) - Michael Koziarski

So this guy is a Rails developer, that is to say he is on the core team developing rails itself. He talked about some of the good and the bad code in Rails and, if I'm being honest, it was slightly out of reach for me. Having only developed one rails app to date my Ruby experience is still limited so when he talked about the good in alias_method I was kinda of confused. I get the idea, I am, by trade, sadly a Java developer and have used AOP so the whole idea of instrumenting a method adding before and/or after code isn't new to me. I did however see and understand the bad stuff. His example opened up ActiveRecord and showed the audience how Rails builds finder SQL, it is nasty. However it works well and is stable, but... it ain't pretty.

The conclusion he arrived at was good. In a nutshell he said that beautiful code is all well and good but ugly stable code that works is also good, because... well... it works and it's stable. He suggested that we, as developers, need to learn when to draw the line. When to say ok I'm done, that does what is supposed to do, now lets get on with some other task. I think that was a great thing to say. As a developers we need to know when to draw the line between beautiful code and code that can be understood, extended and ultimately be refactored to provide more functionality at a future date... but not now.

Pushing Rails onto our Stack: a system integration study. - Richie McMahon & Maria Gutierrez

This was an experience report about introducing Rails into an organisation who's productivity had slowed to a crawl because of, essentially, j2ee bloat. There were a few things about this talk that reminded me of my old pre bank work and I guess that is why it really hit home.

  1. The degree of pride that both Richie and Maria showed at their achievement. I thought that was cool and it made me really buy into the story they were telling. I knew where they were coming from. In the old job I had introduced, after years of banging my fist on the table, some Agile/XP principles; TDD, iterative development and stand up meetings. I felt passionate about that at the time and, I guess, to this day quite proud of what we did in that year.
  2. How Java frameworks slowed productivity. The same project that pioneered Agile methods pioneered the buy vs build approach in my last job. We stopped using home grown frameworks and proprietary code generators and started looking to the open source community. We came up with what had becoming a popular approach of Struts/Spring/Hibernate. Ohhh no!!! I can hear you all say with your convention over configuration hindsight, "that's XML hell". It was and I guess still is. We like Richie and Maria's team had made rapid progress in the early on then slowed under the weight of config... the interesting thing was though this happened over only 12 months.

Edgecase Dialogue - Jim Weirich and Joe O'Brien

So in a nutshell this was about mocking in tests, what should and shouldn't be mocked and what smells that can be identified by using mocks. Not that mocks aren't interesting but, the interesting thing about this was that it was, as the title suggests, a dialogue. You may be thinking OK, I know what you are going to say the dialogue was with the audience, you'd be wrong. The dialogue was between Jim and Joe. They literally acted out a day at work, pairing on some issues that Joe was having with his tests and using flexmock to solve all his problems. So apart from forming the opinion that only Americans would have the balls to pull something like that off and of the existence of flexmock I'm not sure I learned much here, which is surprising because I don't know much and these guys know loads!... we are also jmock crazy at work.

Code Generation: The Safety Scissors of Metaprogramming - Giles Bowkett

This was the highlight of Friday, maybe even the highlight of the whole conference. This guy is funny. Very funny. And clever... the git :) I'm not going to spoil it by giving my own terse summary I'm going to insist you watch it for yourself (or at least a version of it) and, as Giles says, read this book. What I will do is commit myself to a blog entry on code generation. It's a topic which has haunted me for a number of years. I have a love hate relationship with it.

So that was the highlights of Friday, or at least during the day. As expected the next stop after the conference was the pub. The drinking didn't stop till around one and in between the Scotland on Rails guys organised some pizza and a game of werewolf. The pizza was great, werewolf on the other hand... I just don't think it's my bag, saying that I was almost blam'd, and I do tend to be a little self concious when in the presence of uber geeks. Anyway, I did get a chance to be the werewolf and took great pleasure in killing the off Piers Cawley :)


So Saturday stared with a greasy roll and a coffee, which I had to pour out in order to get in a Edinburgh taxi. Great start ehh!!

Domain Specific Languages: Molding Ruby - Joe O'Brien

Joe's talk was great and really explained what a DSL was. Interestingly Koz mentioned that he thought that the term DSL was being abused, not in Joe's talk however. He felt that a lot of so called DSL's he was seeing these days were just good OO code that modelled a domain. Piers added to that saying that code has to be in the language of the domain in order for us as developers to be able to clearly communicate with the customers. As a fan of DDD this was really good to hear. You can see an earlier incarnation of his talk here.

JRuby on Rails: Up And Running! - Charles Oliver Nutter, Thomas Enebo

JRuby rocks. It is of course just Ruby... maybe that's what the 'J' should stand for. But running it on the JVM has so many benefits. But this is why I like it:

  • It brings Rails deployment inside my comfort zone. This may sound like a really lazy thing to say and as I have never actually used fastcgi or Mongrel so I really don't have anything concrete to compare it against. But the thought of using a simple gem to produce a war file then dropping it into Tomcat sounds great to me.
  • Netbeans. Ok as a Java IDE it has for a long time been 3rd behind Intellij and Eclipse but as a Ruby IDE it's the best I've seen. Last year I played with a beta release for a few months and I loved it. It even kicks the pants off of the new Intellij 7 plugin, I'm gutted... I love IntelliJ.
  • Ruby gets compiled to bytecode on the fly when running on the JVM. You can even pre-compile it. This contributes to making it faster, almost 3 times faster, than using the CRuby interpreter. Fortunately, for you, this exceeds my geek threshold by a country mile so I can't even guess how this happens.

So that was it. I had a great time and learned loads, many thanks to Alan, Paul, Abdel and Graeme!

When does registration open for next year guys?

No comments: