Yes, thats what I was finding, the most basic Java stuff at least needs the
core
stream IO API which we don't have in Parrot yet. Have you considered
whether
it would be worthwhile to write a runtime Parrot -> JVM mapper
for classes that aren't translated (ie. any external method reference that
isn't
resolved locally as a Parrot package is proxied through to the real JVM).

This at least seems more doable near term than even making a scratch in
the surface of porting Java APIs.

Concerning objects, I'm sure its too early to start working on Java PMCs
but this
was just a fun exercise for me anyway.

-Melvin Smith

IBM :: Atlanta Innovation Center
[EMAIL PROTECTED] :: 770-835-6984


Melvin Smith sent the following bits through the ether:

> Anyone doing any work with Java bytecode to Parrot?
>
> I've got a java class disassembler 90% working if anyone is interested in
> helping, I'm not sure if this would belong in the Parrot tree or not.

Yes, actually. I've done some initial work but have completely redone
it recently. I gave a talk early on:
http://www.astray.com/java/

I've even got a module on CPAN which does Java bytecode to Perl (for
simple stuff currently):
http://search.cpan.org/search?dist=Java-JVM-Classfile

To be honest, I've been waiting until Parrot can do objects and
classes. Converting the bytecode isn't the problem, it's converting
the Java APIs that is...

HTH, Leon
--
Leon Brocard.............................http://www.astray.com/
Nanoware...............................http://www.nanoware.org/

..... #include <witty.f>



Reply via email to