From: Chris Cummer <chris@...>
Date: 2008-04-05T03:35:17+09:00
Subject: Re: RUBY_IMPLEMENTATION

On 4-Apr-08, at 11:04 AM, Alex Young wrote:

> By using require(), we open the door to being able to select
> implementations in runtimes that are capable of supporting more than  
> one
> implementation of a given feature.  defined? can't do that without
> metatrickery,

Hmm, good point. I'd erroneously looked at your "require 'features/ 
threading/native'" statement as akin to returning a boolean which is  
how Luis' implementation suggestion would work. I see now that's not  
how you'd intended it to be.

And if I now understand you correctly your suggestion also seems to  
satisfy the ability to discern the conditions that Charles brought up  
yesterday:

> JRuby will have native threads, Java libraries, and platform- 
> independence quirks but no continuations or C extension support.  
> Rubinius will have green threads, some C extensions, and  
> continuations. Ruby 1.9 will have native threads with GIL, C  
> extensions, and so on. All three will have slightly different ways  
> to access parse trees, instruction sequences, and runtime internals.  
> All three will hopefully have the same MVM APIs. But it does seem  
> like we're defining characteristics of a RUNTIME more than an  
> IMPLEMENTATION or ENGINE, especially since the definition of what is  
> the "XXX runtime" will probably change while the "XXX engine"  
> remains essentially the same (or in the case of JRuby, the runtime  
> stays the same and there's several engines driving it).

Correct?