Amazing. Reading this reminds me why I _don’t_ like Ruby’s syntax, but nevertheless this takes language integration to new heights I had never seen before. I am quite awed.

Thanks – I actually wrote my 2nd comment before I saw your response. Thanks for the clarification though – it’s much appreciated, your work on this topic is impressive. I hope that JRuby will address this shortcoming in the future. Thanks again!

lol I just get into a rhythm I guess.

@Chris

Yes, it is just the trait mixin functionality, but trait mixins are actually used by a good portion of the library. For example, the wrapped Ruby arrays and Hashes both include RandomAccessSeq and Map, respectively. Likewise, you can use closures, but you cannot pass Ruby blocks as Scala functions unless the method you are calling avoids calling the compose method. Vanilla JRuby allows you to pass blocks as Scala function values, it just has some limitations (outlined in the article).

So to condense my answer: if you want to use the library on JRuby 1.1.4 (or any version with the interface regression), you will need to limit your use to the operator overloading functionality; no trait mixins, no wrapped Array(s), Hash(es) or Proc(s).

The issues are related to the mixins functionality. Generically, any time an interface is involved and a method is overwritten within the Ruby class, JRuby crashes. This was originally identified with two Java interfaces with conflicting signatures, but it seems to also apply to mixins. Quite often (read: all the time) a trait will inherit a member from a super-trait and then redefine it. This is what triggers the crash.

It is possible to just avoid using Scala mixins within JRuby, but then you lose everything except the operator overloading. I guess the only thing we can do is vote for JRUBY-2999 and nag Charlie to get it done.

At the end it’s noted that this does not work with 1.1.3 or 1.1.4 – can you outline which aspects of this do not work? Are there core major issue associated with these releases? or have minor syntactical things changed which have caused some problems?

Seriously though, JRuby is used a lot in Java-based projects which need a more dynamic language for a certain layer (Groovy is used here as well). This works out fine since JRuby has excellent integration with Java, but when trying to use it with a project written in Scala, we run into some difficulties (as illustrated in the first example). This library allows one to use JRuby with Scala and fully leverage its wonderful flexibility. In fact, I would argue that with this library, the JRuby/Scala integration is even closer than JRuby/Java.