I've been doing most of the research on the compilation front, so I'll inject a few points here.- Compilation to an intermediate bytecode would certainly be an improvement. The major refactoring I undertook last fall to change the interpreter from recursive to iterative was an effort toward this en
tis 2006-04-11 klockan 13:41 +0200 skrev Ola Bini:
> Another issue to think about is Ruby 2.0 and Rite. As far as I can
> understand Rite will be a real virtual machine with Ruby byte codes too.
Yeah, that's what I meant by the YARV bytecodes. It's the stuff that
will hopefully be released as Ru
Hi!
Of course, going directly to byte code should be simpler. There is nothing
really comlicated with Java byte code. On the other hand, mapping each
Ruby construct to the specific JRuby Java code would probably be simpler to
map out by first trying it out by generating source code. I'm not
su
Ola Bini wrote:
> It seems to me that a first step for this would be to have the current
> parser output Java source code, which uses JRuby internals to create
> Ruby-classes in the way the builtin classes are defined right now (like
> RubyString and RubyTime, etc). It really doesn't seem that h
Hi,
I'm not familiar with the current implementation, but I how do you implement
it now?
Guy
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] Behalf Of Ola Bini
Sent: Tuesday, April 11, 2006 12:36
To: [email protected]
Subject: [Jruby-devel] JRuby-futu
