Re: State of Perl6 Backends

在 2006/6/22 下午 8:18 時，Swaroop C H 寫到：

> Hi Audrey,
>
> I'm hell confused on the state of the numerous Perl6 backends that  
> we have :-). As I understand it, there is no "main" backend right?  
> First there is the Pugs part written in Haskell that compiles to  
> PIR which Parrot can run, and similarly it compiles to JavaScript /  
> CIL / etc ? Then, we have the misc/pX/Pugs-Compiler-Perl6 which is  
> a p6-to-p5 translator. Is my understanding correct?
That is completely correct, although we don't compile to CIL at the  
moment.

> Going forward, what will be the "main" backend? I mean what will be  
> the one that will be used in production? I am assuming we have all  
> this because of the bootstrapping problem?

The "main" backend as I see it, in the near future, is definitely the  
Perl 5 runtime for production use, because a new runtime (be it GHC  
or Parrot or SpiderMonkey) always faces a far higher barrier of  
entry.  Codegen to that runtime is probably going to be written in  
Perl 5 in Pugs::Compiler::Perl6 space, although it may also happen at  
Perl 6 space, Parrot space, or Haskell space.  (The author-side tools  
has higher flexibility; the client-side runtime must only assume pure  
Perl5 and maybe selected well-known XS modules.)

But in the long run, it'll just be different runtimes for different  
environment.  Python has maybe 10 implementations, half of them quite  
complete; Scheme has more than 20; it's really good for Perl 6 to be  
retargettable across different runtimes as well.
> Sorry for the numerous questions, but I'm so curious, and I didn't  
> know which list to send this to (internals vs compiler vs users ?? )
-compilers is probably it; Cc'ed them. :-)

Thanks,
Audrey

PGP.sig
Description: This is a digitally signed message part