Another year, another RubyConf. 2007 was bigger than last year, maybe with an attendance of 500 people, and it’s hard to say if it’s kept any of its intimacy. The new semi-two-track approach (single-track in the morning, double-track in the afternoons) seems a good middle-ground to me. The only problem is that it’s probably nigh-impossible to predict demand when you’ve got two tracks. There were many afternoon slots in which one of the rooms was packed, the air dense with nerd-musk. Not a good thing.

The conference program overall seemed to have a highly theoretical, philosphical focus: Marcel Molina talked about beauty in code, Luke Kanies talked about Gödel’s incompleteness theorem, and then I talked about post-structuralism and dynamic typing. I was genuinely prepared to have the talk with the least amount of code, but then Luke had none at all, so there you go. Anyway, I enjoyed this aspect of the conference. It’s like we’re becoming a sort of OOPSLA-lite, which is preferable to becoming the next Java One.

I left this conference excited about the year ahead, for lots of different reasons:

• The one-two punch of JRuby, which will tempt more of the enterprise world into Ruby-land, and Rubinius, which will make it easy for programmers to contribute to an implementation without being C wizards.
• RSpec upcoming Story framework, in which high-level tests can get expressed in re-usable, described blocks. I don’t have the problem of needing to make these tests readable and writeable by non-techies, but I sure do have a lot of too-long integration tests in the Rails apps I write.
• Treetop, an easy-looking DSL for text parsing. I don’t know much about parsing and am not sure when I’d use Treetop, but it certainly seemed simple.

I also ended up leaving with a few questions that will have to be explored further:

• Is my sense of testing completely different from that of others? Why is it that nobody seems to feel the same about mocking and stubbing as I do? Why am I fine with writing less longer tests instead of lots of shorter tests? I have a feeling that I’m writing good tests, and that other people are writing good tests, but that there are nonetheless some fundamentally different mental models being employed here.
• What’s up with the resurgence of attempts to transparently map Ruby to SQL? Jim Weirich brought it up during his talk as a learning exercise only, but a later Google search reveals Ambition. Having gone partway down this path years ago with Lafcadio, I’m not sure what to think. Ambition seems to use ParseTree, which would overcome some of the limitations of Lafcadio’s approach. But working with Ruby query inference in Lafcadio made me wonder if the idea’s a good thing at all. More on this later …