fix -keep-hc-file-too

revert rev. 1.81

Typo

Fix obvious typo (pprStgARg for pprStgArg).  Clearly not a frequently-
tickled part of the compiler!

* Improve the warning "M is imported but nothing from it is used"
  In particular, don't warn if some instances from it are imported.

  It's pretty much impossible to do the Right Thing always.
  A comment in Rename.lhs says
	-- NOTE: Consider
	--	      module This
	--		import M ()
	--
	--	 The import M() is not *necessarily* redundant, even if
	-- 	 we suck in no instance decls from M (e.g. it contains
	--	 no instance decls, or This contains no code).  It may be
	--	 that we import M solely to ensure that M's orphan instance
	--	 decls (or those in its imports) are visible to people who
	--	 import This.  Sigh.
	--	 There's really no good way to detect this, so the error message
	--	 in RnEnv.warnUnusedModules is weakened instead

* Minor comment changes to RnIfaces.lhs

* Use NameEnv instead of UFM in TcEnv (tidy up only)

Add Marcin Kowalczyk and Michael Weber to the "Additional
Contributors" section.

A string constant block can consist of multiple .string/.ascii
directives.  The sparc-specific code in this file already handled this
properly, but the x86 code was wrong.

no need to set hsc_lang = HscC when -split-objs is on.

New form of literal: MachLabel, for addresses of labels.  Used by
foreign label instead of MachLitLit now.

Real lit-lits now cause the NCG to panic.

Also: removed CLitLit from AbsCSyn; it was only used in one place for
a purpose it shouldn't have been used for in the first place.

Corrected URLs.

New section on Packages, collecting together the various
package-related documentation and adding some new stuff about the package
management features.

Bump the version no.

ghc 4.06 didn't have -package, so use it only for 4.07+.

bullet-proof the package code a bit more; check for write access to the
configuration file before doing anything, check whether we're adding a
package that's already there, etc.

Packages can now be added/removed from an installed GHC as follows:

    $ ./ghc-inplace --list-packages
    gmp, rts, std, lang, concurrent, data, net, posix, text, util,
    hssource, win32, com, std2
    $ ./ghc-inplace --add-package <newpkg
    Reading package info from stdin... done.
    Saving old package config file... done.
    Writing new package config file... done.
    $ ./ghc-inplace --list-packages
    gmp, rts, std, lang, concurrent, data, net, posix, text, util,
    hssource, win32, com, std2, mypkg
    $ ./ghc-inplace --delete-package mypkg
    Saving old package config file... done.
    Writing new package config file... done.
    $ ./ghc-inplace --list-packages
    gmp, rts, std, lang, concurrent, data, net, posix, text, util,
    hssource, win32, com, std2

This is a first stab at the kind of functionality we need for
installing Haskell libraries via RPMs: the RPM script would install
the libraries, and then do a "ghc --add-package" passing the
appropriate paths.  You'd then have "ghc -package" at your disposal to
use the newly installed package.  Similarly on de-install, the RPM
script would run "ghc --delete-package".

Also in this commit: prettify the package dumping.

Add -t flag to specify title of report.

Rename the marker used by Diff_Gcc_Nat.hs to ___ncg_debug_marker, to make
it clear it has nothing to do with the usual object-splitting machinery.

Improve Diff_Gcc_Nat:
* Emit a warning, and stop, if there are no debug markers in the NCG code.
* Handle .uahalf and .uaword, which appear in sparc assembly, but not x86.

Make -S work when going via the NCG.

From the Ancient Book of Infinite CS Wisdom:
   "Thou shalt remove unused imports after dead code removal..."
;-)

Remove info target for install (we can no longer make info from SGML).

Don't need to install HsStd.h in ghc/includes anymore.  The driver
looks for it here.

Print
   "Interface file unchanged"
and
   "Module version unchanged, but usages differ; hence need new hi file"
only when -ddump-rn-trace flag has been given. This is more consistent
with the rest of GHC and the *nix world, and messes up less tests.

Added semicolon to synch with current interface file syntax.
Note: The `class' line contains a kind bug, but this is not
checked. Bug or feature?

Removed foralls in expected output

Synched expected output with reality

Litlits don't really make much sense for the NCG (i.e. are a hack), so
compile via C.

Reverted my last change to this file: Constant folding should really
yield *excatly* the same result as runtime computation.

Synched with reality again. The expected output is a extremely fragile...

Putting parentheses around expressions involving a mixture of
multiplicative and additive operators might sometimes be a cunning
idea...   :-}   Otherwise
   (591125662431::Int) `div` (517::Int)
is correctly converted into
   (567659506 + (275 * 2147483647)) `div`  517
but
    591125662431       `div` (517::Int)
turns into the messed up
    567659506 + ((275 * 2147483647) `div`  517)

Add note that `gmake clean' should *not* destroy the work of `gmake boot'.
In all cases, the sequence autoconf, ./configure, gmake boot, gmake clean,
gmake all should work correctly.

** Please verify that this is indeed the case! **

DEBUG ==> NCG_DEBUG

D'oh!  Only generate split markers when actually splitting

DEBUG ==> NCG_DEBUG

fix phase ordering for NCG + -split-objs.

Insert split markers properly in the NCG.  -split-objs is on its way
to working with the NCG.

dead code removal

Removed -noC flag, which has vanished from the driver

Don't use addr2Integer for large integral literals anymore, use a
Horner schema with numbers in the Int range instead. This improves
constant folding, so e.g.  (0x87654321 :: Word32) is evaluated at
compile time now. In theory we can completely say Good-bye to
addr2Integer, but for the time being it's still there. Feel free to
nuke it...  >:-)

Don't warn the user about integer overflow during constant folding
anymore. It's not done at runtime either, and compilation of
completely harmless things like

   ((124076834 :: Word32) + (2147483647 :: Word32))

yielded a warning.

Hackily quote -#include options before passing them to mkdependHS