Dave Thomas wrote:

> Yukihiro Matsumoto wrote:
>
>>
>> |As I read this test, I can't help but wish we had a little more 
>> focus on |Rubicon and it was included with the Ruby core.  The 
>> question of "What |should I test?" might have a more obvious answer.  
>> Matz, any thoughts on |putting Rubicon in the core Ruby distribution 
>> (and replacing "make test")?
>>
>> I promised Dave Thomas that rubicon or nearly equivalent test suites
>> will be bundled with 1.8.1.
>
>
> Which has a nice side-effect - we can remove all the version checking 
> in the tests if the test are in the repository alongside Ruby.
>
>

This is great news, Matz (and a nice observation, Dave).  There has been 
a lot of work put into Rubicon, and it would be a shame for us not to 
really make use of it.   It would be wonderful to get it (and Ruby) to 
the point that it could some day pass at 100% on every supported 
platform.  We're pretty far from that goal as of now--especially on Windows.

This would really complement the recent talk of a formal specification 
for Ruby.  We could have an EBNF-ish grammar as a document and and input 
to the parser generator, and Rubicon as an executable check of the 
implemented interpreter.  I'm imagining JRuby, NetRuby, and Cardinal all 
validating themselves against this single, authoritative test suite.

Chad