Getting started with Rubinius development
One of the reasons is the modular design and how Rubinius exposes its internals. Rubinius is well suited for new developers trying to help out, as it takes introspection all the way. Ruby already makes it trivial to inspect a Ruby class with methods such as
obj.methods(shows the methods of
obj). Rubinius goes one step further by giving access to the Abstract Syntax Tree (AST) of the Ruby code, a data structure representing the code. Rubinius uses ParseTree, a representation using symbols and nested lists. For example, this code:
puts "Hello, Rubinius. You rock my world!"is represented like this:
[:fcall, :puts, [:array, [:str, "Hello, Rubinius. You rock my world!", 0]]]If this looks familiar to LISP or Scheme code, that's no accident. The representation is called s-expressions (symbolic expressions), which is also how these languages represent their code.
The benefit of representing an AST this way is that it's easier to read than a data structure made up of linked objects in memory. This becomes even more evident for creating ASTs, for instance for test cases using ASTs.
Reading up on ParseTree can be helpful for working on Rubinius. A couple of tools are already based on ParseTree. Ruby2Ruby takes an ParseTree AST and turns it into Ruby source code - useful for looking at runtime generated Ruby code. Flog is a new tool that analyzes code an generate a quality rating of it. Other tools based on ParseTree, such as Ruby2C or Heckle can provide some insights as well. Another use of ParseTree can be seen in the Ambition, a way of using ParseTree representation of Ruby code to create database and other queries.
Brian Ford discusses Rubinius internals, like getting ParseTree output. Among the topics are tools for looking at ParseTree ASTs for Rubinius code and the underlying bytecode for the code. This allows new Rubinius developers to quickly figure out is going on under the covers or what Ruby source is turned into for execution. More and updated details and command line options can be found in the Rubinius Wiki on Command line options.
For everyone now interested in looking at Rubinius code, Sam Aaron put together a list of links and instructions to get started on Mac OS X. One aspect, that might be problematic for new developers, is the source code management (SCM) system Rubinius uses. After using Subversion for some time, the team recently switched to Git. Git is a Distributed SCM, originally written by Linus Torvalds for Linux Kernel development, but an increasing number of projects are using it too. Distributed SCMs are increasingly popular, as they faciliate branching which allows for easy experimentation
To get started, see Using Git page at the Rubinius website. Once Rubinius is checked out, it's easy to keep up with the development using Git, as Sam Aaron points out:
Head to your Rubinius directory and utter forth the following:
git pull ; rake build
The Linux kernel developers may all be using a certain free UNIX clone, but I bet there are one or two Ruby developers on Windows..
No idea how well it works though.
There seems to be a Java Git client and an Eclipse plugin
It's somewhat surprising (to me at least) that many projects now seem to choose Git, especially since it was designed for Linux Kernel work on Linux. I'd have thought that Mercurial www.selenic.com/mercurial/wiki/
would be the choice for non-Linux Kernel projects. It's used by many projects
and (from what I know) was one of the two final candidates as DSCM for Rubinius.