Refactoring Ruby

August 30, 2006

ArticleS.UncleBob.RefactoringParalysis:

"I've been working on on the Harry Potter problem in Ruby. I have a version that works, and now I'd like to refactor it. I'm using Text Mate as my 'ide'. I find myself in a state of paralysis. There I things I'd like to change, but I know the cost will be high, so I think and think about them instead of acting. This is not how I work in Java with IntelliJ. With IntelliJ I do my thinking by acting! I change variable names; I extract methods; I create and destroy base classes; in short I refactor madly. But in Ruby, without a good refactoring tool, I sit and stare at the code. I know what I have to do. I just have to break through the fear and start refactoring; and I will. But I was amazed at how intense the fear of changing code is when you don't have the appropriate tools. Jetbrains, or someone, you really have to make a refactoring tool for Ruby!!!"

The only apparent project on refactoring for Ruby is this. And it has not been updated for almost a year. Moreover, it uses Emacs which probably means that everything has to be done using lisp macros.

That said, the idea of writing a refactoring browser for Ruby is pretty interesting. And one that I have been thinking of (but not trying to do yet) for almost a year. There aren't that many refactoring browsers for dynamically typed languages. Smalltalk has one (and, mind you, was the first language to have a refactoring browser).

So I am going to gather some more information on this subject and propose this as a master's thesis to Prof. Johnson. After all, being with at this university with the pioneers of refactoring means that there is definitely going to be a lot of people with enough experience. And there is always Don Robert's thesis to help me get started.

I would also like to know the importance of a refactoring tool for a dynamic and agile language such as Ruby. All the latest refactoring tools concentrate on Java and C# both heavyweight languages in their own class. Last summer, I worked on Photran: A plug-in for Eclipse that supports the Fortran language and has some refactoring. Even though I did not work on the refactoring part, I can tell that it was really hard work. Even now, the parser for Fortran is not as fast as most people are used to.

Some possible problems I can definitely think of:

How to get at the syntax tree for Ruby. To make a refactoring browser, I would have to get hold of the syntax tree. I could scheme on this and make it be based on regular expressions but that would make it too rigid. Any slight variation in coding style would probably break it.
Which application to write it in? You have to write it in an application that people are already using. I know that a lot of people use Textmate and Emacs but both those editors do not provide any real tools to help in parsing the code.
What refactorings should be included? I mean there are a couple of standard options but what are the refactorings that most people use in Ruby? Are those refactorings going to be useful say for Rails?
Who would actually use this tool? Is there a market for this? You might argue that Smalltalk was never a major programming language. However, it was pretty big at the time the refactoring browser for it was being developed.
Smalltalk is a pure object-oriented language. Ruby allows you to write things without classes? Would this make a difference? I am not too sure. There is a C refactoring browser written by one of Prof. Johnson's students so refactoring for procedural languages is definitely possible.
Is this even doable for a master's thesis? I am definitely going to try to get into the Ph. D program but if I cannot, I want to make sure that this is doable.

By the way, there definitely is interest in refactoring Ruby on the web: