JRuby Roundup: Java Integration and Debugging (JSR-45) Improvements

by Werner Schuster on Aug 18, 2008 |

The next maintenance release of JRuby is scheduled for release later this month. Among the changes is an overhaul of the Java Integration, which defines how Ruby code interacts with Java libraries. Java Integration concerns basics, such as type coercion or overloading, but also streamlining the interaction between Ruby and Java, eg. making it possible to pass a Ruby Block to a method when it makes sense.

Charles Nutter shows some of the recent improvements in JRuby's Java Integration:

0. Obviously, there's been a lot of performance work.

1. Closures can be passed to any method with an interface as its last argument; the closure will be converted to the target interface type. thread = java.lang.Thread.new { puts 'here' }

2. Interfaces can be implemented using Ruby-cased (underscored) names for all methods now.
class Foo include java.awt.event.ActionListener def action_performed(event) ... end end3. Interfaces with bean-like methods can be implemented using attr*. [..]

4. Interfaces with boolean methods can be implemented using the question-marked version of the method name.

Another improvement is the use of JSR-45 features to improve debugging (Note: link seems to be broken at the time of publication). JSR-45 allows to map source files and source lines to class files. JSR-45 defines class attributes (SourceDebugExtension) that contain metadata in the SMAP format defined in the JSR to define which input source file a class file comes from. A look at the compiler code shows that line number information is already being added to the generated class files. ASM, used for generating the class files, allows to set line numbers via the visitLineNumber method.

With the addition of SMAP files, regular Java debuggers, eg jdb, can step through Ruby code compiled to bytecode (sample pastie showing how to step through a Ruby file using jdb).

The new capabilities are another step towards a fast debugger for JRuby - however it's important to note that this method only works for code that gets compiled to bytecode. Code that is interpreted still needs to be handled by the existing trace or hook based debuggers. How much of an applications code is turned into bytecode depends on a couple of factors. For instance, there's the risk of running out of PermGen space, which usually ends in a JVM crash/termination. To solve that, the JRuby JIT limits the amount of compiled methods (the limit can be configured).