Emitting java bytecode?

[karczma at info.unicaen.fr]

Einar Karttunen writes: 

> Tomasz Zielonka <tomasz.zielonka at gmail.com> writes:
>> For example, some darcs users complain ...

> Perhaps making -keep-hc-files a bit more documented ... 
> 
> This could be faster and easier to implement than creating a version of
> the RTS running under the JVM.
============== 

More generally: do you (Tomasz and others) really think that Java,
or even C/C++ is a good "portable" platform for implementing efficiently
Haskell or other lazy functional languages?? 

I believe this has been discussed already. And GHC passed already
through this stage. I don't believe the world can return to it. The
RT model is different, even decently implemented tail recursion is
not so easy in C, and in Java I wouldn't bet either. 

Moreover, I don't think that Haskell is a good way to perpetuate
imperative languages as "portable platforms". While Lisp machine
belongs to the past, it seems reasonable to hope that "functional
hardware", with some universal implementation paradigms, will be
more popular... 


Jerzy Karczmarczuk