Antonio Cangiano March 4th, 2008
Rails has been a blessing and a curse for the Ruby community. It brought sudden popularity to the language with all the consequences, good and bad, that usually result from exponential growth. On one hand, it gave many developers the chance to appreciate the design of the Ruby language based on its own merit. On the other hand though, it’s been a cash cow that’s changed the community forever by attracting all kinds of attention. Rails has become the poster child for Ruby, blurring the distinction between the Ruby, and Rails, communities. A large number of web programmers got to experience the “ease of use” and beauty of Ruby development, but it also clearly exposed Ruby’s implementation shortcomings. Rails enabled Ruby to go from a relatively unknown programming language to a mainstream one within the frame of just a couple of years. Languages are made to be used, so overall, most would agree that the advantages for Ruby, brought forth by Rails’ success, far outweigh the negative aspects. In the words of Matz, Ruby’s creator:
Rails is the killer app for Ruby.
— Yukihiro Matsumoto
Whatever your take on the subject is, in this article I argue that Ruby on Rails is actually the best thing that ever happened to Python. Rails is a successful Ruby framework, so some may think that it would convince people to switch from Python to Ruby. In other words, naively, one could think of Rails as a Python exterminator, at least as far as web development goes (which is a big deal nowadays). This couldn’t be further from the truth. This year is going to be a great one in terms of the adoption of Python, and I think that Rails has had a positive influence in this regard.
Independently from Rails and Ruby, over the past few years Python has had a good deal of penetration within the market. It wasn’t Java, of course, but as far as dynamic languages go, Python was well represented in many companies and achieved a fairly adopted niche even within the enterprise world. What Rails did was to promote and popularize the usage of MVC web frameworks written in dynamic languages. What could have been considered an unusual choice in the pre-Rails era, is now viewed by most as the right way of developing web applications. The marketing abilities of the Rails community benefited the Python one, in the sense that they made the usage of dynamic languages for serious web projects a very acceptable and even well warranted choice.
Rails got the message out that MVC web frameworks and “scripting” languages could be very productive and much nicer to work with. Django and other web frameworks are indirectly taking advantage of this. A couple of years ago people were asking me whether it was better to adopt Rails or stick with Java for a given project. Nowadays, most emails and requests that I receive are about whether it makes sense to adopt Rails and Ruby or if it would be more sound to use Django and Python.
Initially, even Sun hired Ruby hackers to work on JRuby and have only recently announced that two pythonstas will work on Jython full-time. Yes, they are late to the party and should have done this years ago, but I think that Rails’ popularity and hype led them to consider Ruby first before eventually realizing that they should have done the same thing for Python. Rails has been, at least in part, a catalyst for Python’s success.
Don’t let this confuse you though. Python (and Django) are able to benefit from all the interest geared towards dynamic languages, only because they are technically excellent and make a strong case on their own. Their communities are much less about marketing and more about substance, in my opinion. I understand those who go from Ruby to Python, but there are far fewer motivations in favor of a switch from Python to Ruby. The reason for this is that, in a way, Python is currently an answer to Ruby’s MRI shortcomings. When I speak about Ruby’s shortcomings I always refer to the implementation and not to the design of the language, which is a well balanced and coherent mix of paradigms and features.
Things will change with Rubinius (and perhaps JRuby), but as it stands right now, Ruby’s implementation has critical limitations that affect the development of non-toy projects and its adoption within the enterprise world. Speed, I/O processing, threads, garbage collection and troubling deployment when compared to other languages, are serious issues. It doesn’t mean that you can’t use it, just that the limitations that are in place will make your life more difficult on certain projects.
In a way, Python is the only acceptable implementation of Ruby for certain values of Ruby. The two languages have different approaches and quite a few distinctions which make them both unique. Overall though there are plenty of similarities that make developing in one or the other a somewhat comparable mental process and coding experience. Sure, Python doesn’t promote the functional paradigm as much, and it’s less implicit/magical than Ruby (read import this). This could be a good thing when it comes to large and enterprise projects, but the two are all not that different from each other. They’re Capelli d’angelo and Spaghettini, as my fellow countryman Alex Martelli eloquently put it.
As much as I may like Ruby more as far as language design goes, not only does Python boast a very solid implementation, it has several advantages over Ruby that go beyond the interpreter. I’ve found that Python has an incredible amount of rock solid, high quality libraries that perform very well. Not all of them of course, but most are well coded, maintained and documented. In Rubyland we can’t claim the same levels of good reusable code. I use both and I see a big difference. Lingering for a moment on the subject of documentation, Python has a wealth of tutorials, guides and even entire books available for free online. Learning Python from these, without spending a cent, is a walk in the park. Ruby on the other hand has good books in print, but a long way to go as far as free documentation goes. And this is true for both Ruby and Rails.
The community attitude is much different too. The Python and Django communities generally keep a low profile, following the “shut up and show them the code” mantra. They do have some marketing issues but can’t be blamed for hyping their technology, like at least in part, Rails has done. For example, Django is a very mature framework that’s several years old, and yet it still hasn’t been tagged as a 1.0 version. If Twisted Matrix was implemented in Ruby it would be advertised as the second coming of Christ. 
Jokes aside, I feel that the Python community has a very good attitude which is in no way altered by the Django one, because it happens to share the same traits. There are no exclusive private clubs or the feeling of experiencing a technological “gold rush”, even though the community is in no way smaller. This may seem like a minor point, but for many the maturity, pragmatism and attitude of the community is a big selling point for Python/Django.
I’ll refrain from indulging in a full length comparison of Rails and Django. But I must briefly mention that Django takes the cake when it comes to creating applications that do web publishing (for obvious background reasons). Despite attempts to put Photoshop online, I feel that most of the web is still about publishing and interacting with published data. This makes Django a good choice in most cases. Of course, should you develop a web application that intends to replace a desktop app, Rails will most likely have the edge there.
So what does this mean for me personally? I’ll use them both, as I’m a firm believer in using the right tool for the right job. I’ll give you an example. Stacktrace.it is a small revolution in the Italian IT publishing world. A few months ago I contacted about 30 people from amongst the best Italian hackers and IT professionals and I proposed that together we create a site in which we’d promote and influence software development and IT in general in Italy (even if I live in Canada myself). The site has been going very well and has attracted a group of highly technical people who follow and contribute regularly with high quality articles. The code was based on Luambo, an open source blog engine yet to be officially packaged and promoted to the world. What was remarkable about it was that most of the new features that were requested in our private mailing list ended up in the code in less than five minutes. An unbelievable level of productivity. Even our designer, who was not familiar with the language or the framework, was easily able to work on the project. Guess what? Despite loving Ruby and Rails, when it came to deciding on the framework and language to be used for that project, I strongly pushed for the adoption of Django and Python, and not Ruby on Rails. It was simply the right tool.
