Gave driver access to bash for running system calls under Windows by writing
command out to a temp file, then sending that as the argument to "sh -". Sigh.
This is #ifdefed on mingw32_TARGET_OS; saner OSs just use system as normal.

Stop doing all system calls via "sh -c", as this messed up quoting.

oops, ghc-inplace ain't an EXE.

Added a few EXE_SUFFIXes to get the driver going on Windows.

Changed ExeSuffix to exeext (which is what configure.in actually calls it).

remove hscpp

Add $(EXE_SUFFIX) to hp2ps and unlit

Move dlls to $(DLL_PEN) after generation.

Move DLLs to $(DLL_PEN) after generation.

Arrange that type signatures that have no "parent" signatures
work right.  For example

	   module A
	      f :: Int -> Int

The panic in RnEnv.lookupBndrRn becomes an error message.  Also
RnBinds.rnTopMonoBinds and rnMonoBinds were wrongly ignoring all
signatures altogether if there were no bindings!

fix finding of the usage message (again).

Add comment

Add type-sig test

Holding pen for dlls to stop them getting in the way of pre-compiled
dynamic programs.

Remove dead code: checkConstraints and univErr

Insert code to handle instruction counts from Heimdall, as well as
CacheProf.

Fixed a bracket-o.

Windows fixes and tidying up.

Windows fixes.

Changed __MINGW32__ to mingw32_TARGET_OS to cope with latest mingwin.

Rename files in line with RTS.

Use GMP dll rather than static library under mingwin.

Removed definitions of symbols which are now defined in the mingwin header
files.

Fix bug in SetLevels that meant an occurrence of a non-cloned binder,
where the non-cloned binder shadowed a cloned binder, resulted in
the occurrence incorrectly being substituted with the clone of the
*outer* binder.  Curious that this never got tickled before!

Eliminate need for built-in Rational type

Print operator names in HsExpr better

Pass flags -agc to $(HAPPY) if the $(HAPPY_VERSION) is >= 1.8
(previous versions weren't reliable enough with these settings).

Merged Julian's changes to make it more comprehensible to non FP programmers.

accept '-fvia-c' as an alternative capitalisation for '-fvia-C' (the
old driver used to do this, and it's documented in the Users' Guide)

Add comments

Document new -fstrictfp option

This commit tries to fix the discrepancies between the results of
floating point calculations during runtime and compile time, see
e.g. fptools/ghc/tests/numeric/should_run/arith008.hs. Part of the
story was the fact that floating point values are represented as
Rationals in GHC and therefore never lost precision, at least for the
operations for which constant folding is done. To compensate for this,
before and after floating point operations the operands are
temporarily converted to/from Float/Double. This is wrong, because
host architecture and target architecture are confused this way, but
in a non-cross-compiling scenario it works.

Added new -fstrictfp flag

* Fixed handling of filename arguments without extension
* Added new -fstrictfp flag
* Some cosmetic changes (unless/when instead of if)

Merged Marcin's patches for unboxed thingies

Functional Dependencies were not getting simplified away when the dictionary
that generated them was simplified by instance resolution.  Fixed.

Add mutually recursive kind inference test

Fix typeKind

Arrange that type signatures work right.  Consider:

	   module A
		import M( f )
		f :: Int -> Int
		f x = x

Here, the 'f' in the signature isn't ambiguous; it
refers to the locally defined f.  (This isn't clear in
the Haskell 98 report, but it will be.)

Wibbles in the new kind-checking stuff