> I would like to know that anything I do in hugs is portable to GHC.
> It seems like the two main big differences between the two
> are TREX and the FFI (is there more?).
> Is there a plan for GHC to support TREX?
> What is the status of support for the new FFI?
Yes, the FFI will be supported by Hugs; quite when I'm not sure.
'foreign import' is already supported in the (unreleased) STG-Hugs.
We plan to implement 'foreign export dynamic' too. Bottom line: GHC
and Hugs will agree about FFI in months, not years.
I too would like TREX in GHC. But first I'd like to think more about the
language design aspects -- it's ugly to have two different record
systems in the same language. Mark J and I have just started a
low-bandwidth discussion about this. If others are interested in
joining in, let me know. Bottom line: language design first,
hence longer timescale.
> Also, and more blue sky, is there a plan to more directly support
> the more advanced stuff posted to this list:
> * haskell preprocesor (Derive or PolyP)
> * dependent types or dynamic types
> * arrows vs monads (or library cleanup more generically)
> * assertions
> * haskell vm (binary interface spec)
I know of no plans for any of these (except for the Hugs/GHC Dynamic
library; and GHC does support assertions in an undocumented feature).
But the field is open! There are a lot of bright people
on the Haskell mailing list. The most creative
work is coming up with concrete designs --- and anyone can do that.
Simon
