Re: Problem with 4.03 FFI on Sparc?

[Manuel M. T. Chakravarty]

Sven Panne <[EMAIL PROTECTED]> wrote,

> [ redirected to ghu ]
> 
> "Frank A. Christoph" wrote:
> > I wrote:
> > >  * A more pressing point is that GHC is tied to x86 machines at
> > >    moment, see e.g. MBlock.c or Adjustor.c.
> > It is? I thought these were only relevant for the FFI.
> 
> OK, I was a little vague: MBlock.c is needed by the RTS in any case
> and *guesses* a free region in virtual memory for most architectures.
> Most guesses seem to be quite good, though. Adjustor.c is only
> necessary for foreign export dynamic, so most programs (if not
> written by Manuel or me :-) won't notice the omission, but it is
> exactly this kind of tiny differences between platforms SuSE tries
> to avoid.

Furthermore, IMHO, the lack of a working FFI seriously
compromises Haskell's value as a general purpose programming
language.  So, I can understand SuSE insisting on this
point.

Manuel

PS: An intended implication of this comment is that I
    believe that the Haskell standard should include a FFI.