GHC 5.00 release update

[Manuel M. T. Chakravarty]

Simon Marlow <simonmar@microsoft.com> wrote,

> Just to keep you up to date on where we're at.  We're planning to enter
> Feature Freeze for GHC 5.00 within a week (two at the
> most).  

Whoooohooo ;-)

> Before
> then, we're hoping to get GHCi working on Windows, and implement
> do-notation on the GHCi command line.  If anyone has any more
> *essential* feature requests, please jump in now.  

One of two things (or best both):

* Remove the `with' of implicit parameters in favour of `?x =
  ...' bindings and/or

* don't lump the FFI in with the rest of -fglasgow-exts and
  have a -fffi or so.

I really would like to have the latter.  Currently, just to
get the FFI, I, eg, also have to allow overlapping instance
declarations, which makes it harder to write portable code
(in the sense of the compiler doesn't spot non-conformance).

I have already looked at removing `with'.  It doesn't seem
hard, but there is a nasty quirk.  We need to allow
`... where ?x = ...' and convert it to the `HsWith' in the
abstract tree, but this is complicated by the fact that
`where' can be used to bind over multiple bindings, which
`HsWith' can't represent that easily.

It shouldn't be hard to add a `-fffi', I guess (haven't
looked at it yet).  I am happy to do it myself if we agree
that it is a Good Thing.  (-fglasgow-exts may still imply
-fffi, but I want to be able to switch the FFI on
individually.)

Cheers,
Manuel