Simpltry

RailsConf 08 day 2 was much better.

Jeremy Kemper gave the morning Keynote. It was about all the new shiny in rails 2.1, which is released today. He went over active_record scopes, migrations, time zones, and other things. I think the changes to migrations are brilliant and also painfully over due.

The first session I went to was on edge caching and ESI. It was given by Aaron Batalion. ESI is an alternative to fragment caching that is heavily used by akamai. I really liked this talk, it went at a good pace and the speaker was well informed.

The second session I went to was Advanced RESTful Rails by Ben Scofield. Unfortunately, my day job isn’t doing anything in a restful way. All that a side, I LOVED THIS TALK. I will definitely investigate how to start incorporating REST into my day to day.

The next session was Fast, Sexy, and Svelte: Our Kind of Rails Testing Dan Manges (ThoughtWorks), Zak Tamsen. They had some interesting ideas, and I really appreciate deep test. But, I don’t agree with unplugging your unit tests from the database which was one of their major points. I agree with their goals of making tests run faster and me more informative. I just don’t think the effort justifies the value. So, I will continue to have terrible average tests.

The next session was Integration Testing with RSpec’s Story Runner by David Chelimsky. I tried to go to a different talk, but I got there after the room filled up, so what I am about to say is as a person who didn’t want to go. RSpec Story Runner seems like an extreme pain in the ass. It is like why use one step when 100 will do. David was very well informed and well spoken, I like his style but not the content.

The last session, The Great Test Framework Dance-off by Josh Susser. I love his blog, and he is a great speaker with a very objective view point. And that is why he came to the conclusion that the best test framework is a matter of opinion. No surprises there!

And the Keynote by Kent Beck was a series of stories. About the “creation” of TDD, XP, and Design Patterns. It was very introspective and personal. He mentioned on several occasions the “watering down” of XP into “agile”. In his words, “who wouldn’t want to be agile”. I completely agree with that. The word agile is essentially meaningless. He came up with a new term, “responsible programming”. That basically refers to transparency and quality in the development process. I couldn’t agree more with the notion, but the phrase feels a little watered down (which I believe is the point.) So that makes an essentially good idea, 0 for 3 in naming in my opinion, extreme, agile and responsible.

I have so much respect for Kent Beck. He is a very humble and thoughtful person. I appreciated his talk more than anything at the conference. This is especially amazing considering it will have almost no practical impact on my day to day life.

Recapping the official start of rails conf.

The morning Keynote was by spolsky. He was entertaining, but mostly content free.

Dan Benjamin gave a talk called entrepreneurs on rails, I thought it would have something to do with rails, my mistake. It was still a good talk about making decisions to market and promote in addition to code.

Next, I went to a talk called the hosting woes, by the engine yard peeps. It was not that organized, more of a panel style discussion. They had some interesting things to share about how to structure your server environment. Like, put a few mongrels and a memcached instance on each “app” server, because that will be the most effective use of cpu and it scales more naturally.

Then I accidentally ended up in the DataMapper. I like DataMapper way more than I thought I would. I still don’t know if I have a place for it in my life right now, but I will definitely evaluating it when I start a new rails project.

The rubinious talk was awesome. I think in 3 - 6 months it will be production ready. One very cool thing they did was implement a version of eval in ruby in like 12 lines of code. It mad me feel like I could contribute, which I believe was the point.

DHH keynoted about, well, I am not sure. I guess it was about “the surplus”, which refers to the productivity “surplus” we have in the rails community. Mostly about how either rails will become “mainstream” or it will be passed by something else. Then the second part of his talk was about using that “surplus” to become a better programmer. That way if/when the good times end you will still have a competitive advantage. Some of his assertions were you should, sleep more, do things not in programming, read books not tech related, programming less, start something from scratch, start sharing, and spread the good word of rails. Mostly I think, if you read signal vs. noise, it was kinda obvious. Still, it was good to here it reiterated. Although, it kinda seemed like a sermon, mostly information you already know, but when you hear it, it makes you happy. But mad props for using the phrase, “go dubai”.

Again, more to come later, but if you want less information more frequently, you can follow me on twitter @schlu.

and/or subscribe to my RailsConf ‘08 flickr stream.

I am currently in Portland at RailsConf ‘08. The first day went well.

We did 4 hours on Meta programming which is enough to make anyone’s head spin. There were a few too many matrix and star wars references. I know we are geeks and everything, but it almost became a parody on geeks.

Then I went to something from the Relevance crew called Refactotum. It went over some cool tools. Some of which I use, rcov. Some of which I have heard of, flog. And some of which I haven’t, tarantula. These are all testing tools to help you find problem areas of your code.

The main part of Refactotum started in the middle. We all sat down for a while and contributed patches back to open source projects of our choosing. Before we got to that though, they started it out with the question, “how do you know what to contribute?” The answer was, look at the “problem” areas of the code with the tools above.

More to come later, but if you want less information more frequently, you can follow me on twitter @schlu.

and/or subscribe to my RailsConf ‘08 flickr stream.

As our project gets larger and larger at RideCharge, I have become more reliant on our test suite. We don’t do anything fancy, just use the de facto rails Test::Unit. I have never been serious about BDD, but I think I would like it.

I have flirted with and really like the idea of rspec, but on a project where you have a large existing test suite Shoulda is a drop in enhancement. It’s biggest win is that it allows you to give your tests more context. But, there is a secondary reason for using Shoulda, it allows you to follow the 1 test 1 assertion rule without breaking a sweat.

example:

context "a user that is in pending state" do
  setup {@user = User.find(:first, :conditions => {:state => "pending"})}
  should "not be active" do
    assert !@user.active?
  end
end

You can just put this code right in your User unit test without making any other changes. When the test fails it will read, a user that is in pending state should not be active. That is more readable, especially when the build breaks and you have to dig though a thousand assertions to figure out which one is the culprit.