Add a section describing GHC's version numbering policy.

wibble

- add a new section on "keeping intermediate files"
- document the new -keep-tmp-files and -keep-raw-s-file options.

Add -keep-tmp-files flag, to tell the driver not to delete tmp files.
Useful mainly for debugging the driver, and obsoletes the GHC
folklore of using ^Z for this purpose.

This commit completely re-does the kind-inference mechanism.
Previously it was inter-wound with type inference, but that was
always hard to understand, and it finally broke when we started
checking for ambiguity when type-checking a type signature (details
irrelevant).

So now kind inference is more clearly separated, so that it never
takes place at the same time as type inference.  The biggest change
is in TcTyClsDecls, which does the kind inference for a group of
type and class declarations.  It now contains comments to explain
how it all works.

There are also comments in TypeRep which describes the slightly
tricky way in which we deal with the fact that kind 'type' (written
'*') actually has 'boxed type' and 'unboxed type' as sub-kinds.
The whole thing is a bit of a hack, because we don't really have 
sub-kinding, but it's less of a hack than before.

A lot of general tidying up happened at the same time.
In particular, I removed some dead code here and there

Remove dead code

Take the README for a binary dist from the project directory, not fptools.

update, and make less GHC-specific

update to match reality; try to be a bit more helpful

Zap yet another bogon in x86 floating args to ccalls.

more changes to FPTOOLS_GHC_VERSION:
* fixed bug in too-many-args case
* reworked regexp for versions

sparc NCG is now enabled by default for non-opt compilation.

Reg needs to be imported non-abstractly for sparc.

Make the x86 NCG work again following recent sparc hackage.

Also, fix the x86 bits pertaining to the floats-promoted-to-doubles-
in-ccalls problem.  So this problem should no longer exist on x86
or sparc via NCG.

use Maybe.isJust instead of proprietary verison.

added Debian location to DocBook catalog test

typo

updated comments

Hash MachLabel, too

Replace RcFloating with RcFloat/RcDouble. I'm not exactly sure what
I'm doing here, so somebody else should better have a look.

* linux_TARGET_ARCH => i386_TARGET_ARCH
* move callClobberedRegs into safety in front of the register
  #define-orgy (what the ... is this cpp-ery for? %-]

reinstate used import

remove unused imports

remove unused imports

remove unused imports

remove unused imports; misc cleanup

remove unused imports; misc cleanup

remove unused imports

use MachLabel rather than MachLitLit for compiling foreign label and
foreign export dynamic.

fix typo

undo what appears to be an accidental commit

Fix up the sparc native code generator.  Mostly dull stuff.  Notable
changes:

* Cleaned up ccall mechanism for sparc somewhat.

* Rearranged assignment of sparc floating point registers (includes/MachRegs.h)
  so the NCG's register allocator can handle the double-single pairing
  issue without modification.  Split VirtualRegF into VirtualRegF and
  VirtualRegD, and split RcFloating into RcFloat and RcDouble.  Net effect
  is that there are now three register classes -- int, float and double,
  and we pretend that sparc has some float and some double real regs.

* (A fix for all platforms): propagate MachFloats through as StFloats,
  not StDoubles.  Amazingly, until now literal floats had been converted
  to and treated as doubles, including in ccalls.

Test completed, file deleted.

New autoconf tests for ghc version from Michael Weber, with minor
modifications by me.

workaround for strange gcc-ism: elide 'popl %edx' in the epilogue.

Test file; will shortly be removed.

Removed carriage returns (\r) from source files. Please don't check in such
things; they can cause problems on Cygwin (funnily enough). I'm looking into
how to avoid commiting carriage returns when working under Windows.

Third attempt by two people trying to get a simple fprintf right
again...   :-}

Install HsStd.h, stgio.h, stgerror.h, and fileObject.h, too. Hmmm,
this looks not particularly nice, perhaps we should merge those into a
single HsStd.h some day.

By definition (sec. 7.3 in the H98 report), userError raises an
IOError, so Prelude.catch should catch UserError exceptions, too.
Apart from not conforming to the report, without this patch GHC's
typechecker falls straight on its face.