An Interview with Obie Fernandez on Rails 3
Russ Olsen: First, congratulations on the upcoming publication of The Rails 3 Way. The Rails 3 Way is actually a new edition of your 2007 book The Rails Way. I notice that virtually the first thing that you talk about in The Rails 3 Way is Bundler, which didn't exist when you were writing the first version of the book. What do you think are the most important developments in the Rails world in the past few years?
Obie Fernandez: Well, maybe it's the obvious answer, but I can't think of anything more important that the merge with Merb, resulting in Rails 3. That event cemented the position of Rails as the one dominant web framework. It reverberated across the plains of open source as proof that competing frameworks can cooperate and thrive. I think it also gave the community the freedom to invest more heavily in pushing Rails forward without worrying that it might be a waste of time. I never bought the technical reasons to hate on Rails—I considered them a push towards premature optimization. What we're seeing now is that you can squeeze a ton of performance out of Rails without sacrificing the ease of development that made it such a joy to use in the first place.
Russ: One of the things that is new in Rails 3 is the fact that it comes packaged in what seems like 500 separate gems. Do you think that this is a good idea?
Obie: Yes, as long as it's properly managed, and Bundler seems to do a good job with that. Modular, de-coupled design is a great thing and a big improvement from the mishmash we had before.
Russ: Is there anything in Rails 3 that you would have done differently?
Obie: I would have liked to have seen a little more thought put into ActionView. The rendering system, particularly around layouts and partials, is still very tricky and full of opportunities for improvement. It would have been nice to switch the entire test suite over to RSpec too, but I realize that's never going to happen. The existing Test::Unit-based tests are good enough.
Russ: Did you discover anything that you didn't know about Rails in the process of doing the second edition?
Obie: Too much to describe here properly. Rails has a very dynamic community, and the original edition of The Rails Way was finished in fall 2007. That's 3 years of advancements, during which I spent a lot of time growing my business instead of heads-down coding. There's just so much I learned by exploring and explaining the new stuff. ARel, Active Model, and Devise are just a few examples, and the list really could go on and on.
Russ: I'm still spending much of my time working on a largish Rails 2.X.X application. Do you have any advice for folks like me when we finally come to porting to Rails 3?
Obie: My strongest practical advice is to have a comprehensive integration suite in place before attempting to upgrade. Also, upgrade one aspect of your application at a time. For instance, start with Rails 3/Ruby 1.8.7, then go to Ruby 1.9.2. If you're upgrading to Capybara 0.4.0, leave that until after the Rails/Ruby upgrades. That particular leap is a beast of its own that has eaten my last 3 days.
Russ: The original Rails Way topped the scales at 912 pages. How did you approach the task of reworking such a large book?
Obie: Iteratively. I knew the whole book would need to be revised. What complicated matters is that I had to start the work much before Rails 3 was completely defined and ready.
The first step was to get the book's source material into a usable format. The first edition was authored in Microsoft Word, and I was not eager to repeat that horrible experience. Thanks to some crucial help from Eliza Brock and others, we were able to take MS Word output and convert it into OpenOffice XML and then use XSLT to turn that into LaTeX source files and store the entire project on github.
Once the source material was text-based and in source control, collaborating with others was much easier. I started hosting "TR3W sessions" in my living room, where Tim Pope, Jon Larkowski and other guests would hang out and help me go through the arduous process of the rewrite. I'd plug my MacBook into my big-screen TV and mirror the screen so that the other folks in the room could see what I was doing as I did it. I'd ask, "How does this read?" and people would give their comments and suggestions.
Tim was particularly helpful because he's such a speed demon with code. I'd point to a particular code sample and ask if it still worked, and almost before I got the words out of my mouth he had typed it in and executed it and verified its operation. His encyclopedic knowledge of Ruby and Rails was indispensable and saved the project hours of tedious research.
The revision process proceeded chapter by chapter, but we did jump around a fair bit to accommodate our moods. There's only so many weeks you can spend dissecting the finer bits of Active Record behavior before you want to jump into a lighter, easier topic such as caching or authorization. The little wins filled in the gaps between the bigger wins, and eventually most of the book had been revised.
Russ: I notice that you rearranged a fair bit of the content in the new edition, particularly in the early part of the book. Did you do this because of changes to Rails, reader feedback on the first edition or some other reason?
Obie: I've always wanted the book material to build on itself along the lines of how a Rails process bootstraps and handles a request. So Bundler and configuration settings had to come first. But for the second edition, I decided that it was important to discuss routing and REST before diving into the dispatcher. That was the main change in the early part of the book.
Russ: Do you have any other literary projects in the works? Any plans for a third edition?
Obie: No plans for a third edition yet. As for my grander literary ambitions, the answer is yes. I would like my career to point me in the direction of being able to write full-time well into my older years. Jerry Weinberg and Martin Fowler are two of my heroes in that respect, and to follow their lead, I can't stop writing now. I have a manuscript that is over two-thirds finished about the business of custom web development that I'll give attention to again in the coming months. Projects in earlier stages of development have to do with careers in web programming and other topics that I hope will give me exposure to an audience beyond the Ruby community.
Russ: Along with being an author, you are also the editor of Addison-Wesley's Professional Ruby series. Can you tell us about any other upcoming Ruby or Rails books in the series?
Obie: Well, it's an exciting time to be a fan of this series. Michael Hartl's Ruby on Rails Tutorial is selling like hotcakes and garnering a ton of critical praise. Tammer and Chad's AntiPatterns book is also selling well and helping Rails people of all stripes avoid (or fix) the kinds of mistakes that we've painfully learned over the last few years. As for upcoming titles, your new book Eloquent Ruby should be out early next year and will help our readers mature their understanding of Ruby idioms and style. We also have an exciting title in early development that is sure to be a winner: Durran Jordan, author of Mongoid, is working on a hard-hitting new book NoSQL The Ruby Way. He's got serious hands-on knowledge with production use of NoSQL in complex, high-scaling environments. I'm sure our readers appreciate the extent to which our authors are practitioners of their craft.
Russ: You are also the founder and guiding light of Hashrocket. What have you all been up to down there in sunny Florida?
Obie: Thanks for asking. We've been growing—not just in numbers, but also in the arts and discipline of software craftsmanship. I've got a great bunch of people with talent and enthusiasm for taking care of our clients like none I've ever seen before. It really makes me beam with pride to talk about our "Rocketeers."