Fixes for MVars.

Don't bomb out with "Dont know where to get memory from on this
architecture" on Win32.

'Native' Wni32 implementation of getMBlock(). Also added a Win32 impl.
of freeMBlock() in case that ever should become useful.

Documented isEmptyMVar

Documented IO <--> ST coercion operators

Be a little bit more user-friendly when encountering record puns,
which is no longer supported by Haskell. Instead of a Plain Old Parse
Error, print out

  foo.hs:11:11: Haskell 98 does not support 'punning' on records ....

Hopefully that should give the user a big enough hint at what's
going wrong..

Compile the RTS with -O2 and -fomit-frame-pointer.

Added options that 3.xx supported for dumping out selected
chunks of the .hi file to stdout, i.e.,

   -hi-with-imports      = prints out the import 'section' of the .hi file
   -hi-with-declarations = print  out just decls

(Only) useful when working with runstdtest, where we just want
to look at selected portions of the generated interface file.

- Add new object BLACKHOLE_BQ: now a BLACKHOLE is defined as having an
  empty blocking queue, and the first time a thread blocks on a BLACKHOLE
  it is changed into a BLACKHOLE_BQ.

- Remove UPD_INPLACE1 and replace it with UPD_IND in the two places it
  was used.  UPD_INPLACE1 wouldn't have worked in a generational setting.

Re-integrated mod. that seems to have been dropped on the
floor when new-rts moved back onto the main trunk. Here's
the commit msg. that was originally used:

  Extend hConnectTo to also allow output handles to be connected, i.e.,

    h1 <- openFile "foo" WriteMode
    h2 <- openFile "bar" WriteMode
    hConnectTo h1 h2

  will cause h1's buffer to be flushed when h2's buffer overflows
  (and it is just about to be flushed.) The implementation is currently
  not as lazy as that, it flushes h1's buffer regardless of whether a
  write to h2 causes h2's buffer to overflow or not.

  This is used to connect 'stderr' and 'stdout', i.e., output on
  'stderr' will now cause 'stdout' output to (first) be flushed.

- __HASKELL1__ is now 5 (was 4).  Surely not the last word in
  Haskell98 CPP symbols.

- Fix stats munging for new RTS.

Forgot a couple of STGCALLs.

Annotate the 'foreign import's with 'unsafe'.

Tweaked to accept empty declarations (again.)

Haskell 98 updates.

Remove punning.

#if __HASKELL1__ > 4
import Ratio (numerator, denominator)
#endif

Remove Void use.

Parser didn't like "instance Foo X where {}". Does now.

wibble

fix isAlphanum/isAlphaNum problems.

A couple of Int type signatures to avoid H98 Integer defaulting.

H98 changes for non-cygwin targets.

Remove 'const' modifier from static closure declarations.  Static
closures are meant to live in R/W data space.

Minor syntactic h98 updates

Recognise "-- !!!" (as well as "--!!!") as the start of an interesting
comment. The latter isn't liked by a Haskell98 tokeniser, hence the change.

Removed debugging code that accidentally was left behind.

lexLitChar: '\DEL' wasn't being lexed properly

Fixes to support .hi unfoldings containing "_ccall_ dynamic"s

Tweaked to cope with Haskell98's change to what constitutes a -- comment

added isEmptyMVarZh macro

Forgot to commit this before Christmas...on Win32 platforms use
VirtualAlloc() instead CreateFileMapping()&friends to ask for
mega-blocks from the OS. Windows 95-friendlier.

Haskell 98 related changes + the addition of the MVar primop,

   isEmptyMVar :: MVar a -> IO Bool
     -- use with care.

-syslib posix gets the Haskell 98 treatment

-syslib misc meets Haskell 98.

* Misc changes to reflect that we're now speaking Haskell 98.
* Augmented IOExts export list with

		unsafeIOToST	   :: IO a   -> ST s a
		stToIO	           :: ST s a -> IO a

Changes to make the Prelude comply with Haskell 98.

I claim that this completes GHC's implementation of Haskell 98 (at
least feature-wise, but there's bound to be some bugs lurking..)

Minor Makefile mods.

Assorted minor Haskell 98 changes:

  * Maximal munch rule for "--" comments
  * _ as lower-case letter, "_" is a reserved id. Prefixing unused
    variable names in patterns with '_' causes the renamer not to
    report such names as being unused.
  * allow empty decls
  * comprehensions are now list comprehensions, not monadic.
  * use Monad.fail to signal pattern matching errors within
    do expressions.
  * remove record punning.
  * empty contexts are now legal  (go wild!)
  * allow records with no fields
  * allow newtypes with a labelled field
  * default default is now (Integer, Double)
  * turn off defaulting mechanism for args & res to a _ccall_.
  * allow LHSs of the form  (a -.- b) x = ...
  * Main.main can now have type (IO a)
  * nuked Void (and its use in the compiler sources.)
  * deriving machinery for Enum now also generate 'succ' and 'pred'
    method bindings.

x86: only add -fomit-frame-pointer when compiling .hc files.  It gets
in the way of debugging for normal .c files.