Sunday, April 26, 2009

ACCU conference 1 of many (Wednesday)

Another year, another great ACCU conference. I counted 9 languages on the conference program, everything from Scala to C++ - yes there was more C++ that anything else but it was far from the only thing going on. Languages plus sessions on design, pattern, management, Agile and neuroscience.

Normally some of the most insightful comments come not in the scheduled sessions but in the discussions over coffee and in the bar afterwards. Sometimes you don’t know these are insightful until after the event when you’ve left and had time to think it all through.

This will always be a special ACCU conference for me because I was the keynote speaker on Saturday. My keynote was well received and the slides ....

So here is a brain dump.

The opening keynote on Wednesday was from Bob Martin - “Uncle Bob” - about software craftsmanship. (I’m watching the software craftsmanship movement with interest and will write a blog about it in future.) Bob suggested that the most successful Agile method - Scrum - was the accidental “winner” of the Snowbird summit in 2001. This was because the Agile manifesto and Agile values said nothing about technical practices.

Scrum, as is well know, does not contain any technical practices. Teams often borrow these from XP. Because of this Bob believes Scrum is a subset of XP - a subset which just deals with project management. Without the technical practices productivity falls off and teams make a “big mess faster.”

Bob’s answer is a renewed emphasis on technical practices - hence software craftsmanship.

Bob is a great showman and his talk was very entraining, if you get the chance to see him do so. Along the way made several Scrum jokes and was quite critical of Scrum Master Certification.

Someone like Bob can do this, Bob has the reputation and standing to be able to take pot shots from a safe(ish) position. This is good because someone needs act as a foil.

Jutta Eckstein’s session on transparency in Agile was interesting. It reminded me of the session I ran here a few years ago called “Exposing problems / Creating awareness.” The these was similar: “Agile as a problem detector.” The session was largely a discussion between Jutta and the various people in the room.

Keith Braithwaite presented the latest instalment of his “Measuring the effects of TDD” research project. I’ve seen Keith talk about this subject before and over time his presentation is getting more and more compelling. His sample size in increasing and its looking like a more convincing case.

If you don’t know the work have a look at Keith’s blog. To summarise it: you can measure code complexity using a measure called cyclomatic complexity. Keith’s research shows there is a correlation between complexity and automated until tests. Namely: code with a high level of automated unit tests exhibits lower complexity metrics and people comment they are “happier” with the code.

Having discovered the metric Keith is still trying to understand the cause-and-effect relationship and, perhaps, more importantly, uncover how this relationship can be used to improve matters.

Nicolai Josuttis finished the day with a second keynote. You could call it the antidote to software craftsmanship: Crappy code.

Although Nico made it jovially it had a serious point: there is a lot of bad code out there and it is likely to be around for a while to come. The economics of software development and business tend to perpetuate this situation. Therefore, we need to get used to crappy code and find ways of dealing with it.

While I agree with Nico I can’t help feeling it was a little defeatist. If we accept this poor state of affairs do we still have hope of a better tomorrow?

Between them Bob and Nico certainly set a debate raging - Good code, or Crappy code?

Wednesday finished with 40 of us in Chutney’s curry house.