#warning is not ANSI.

Work on initialisation of persistent compiler state

mkSysModuleFS -> mkSysModuleNameFS

Simple fix for install of HTML: mimic recursive install with chmod and cp.

isUnboundName moved

mkWiredInName fixes

FastInt fixes

FastInt fixes; remove unused imports

FastInt fixes

RdrName->Name fix

More of Simon

Removed out-of-date Windows gumph. It's all in the installation notes, ad
naus.

Remove wired-in names.  Partially propogated.

update FAST_INT names

Simons work, mainly on the type checker

Move FAST_INT and FAST_BOOL into their own module FastTypes, replacing
the macro definitions in HsVersions.h with real definitions.  Change
most of the names in the process.

Now we don't get bogus imports of GlaExts all over the place, and
-fwarn-unused-imports is less noisy.

Simons work, mainly on the type checker

Most _RDR names are now proper Names.

The knownKeyRdrNames mapping from RdrNames to Uniques is now just a
list of knownKeyNames.

This change hasn't been propogated yet.

Julians progress

mkSrc* -> mk* (forgotten in last commit)

Rename a bunch of mkSrc* things into mk*'s.

Propagate recent changes in module/Module.lhs.

Add ModuleEnv stuff.

Changes to do with making Module and ModuleName into Uniquables.

Rationalise Module/PackageInfo/ModLocation mess.  Now we just have
Module and ModuleKind.

TLA renaming.

DynFlag plumbing.

we're now at version 4.11, at Manuel's request.  The next release will
be version 5.00 (yay!), and version 4.10 will never exist.

Do most of the DynFlags plumbing.  Also remove stuff pertaining to
search paths since the finder does all that now.

Beginnings of renamer and typechecker stuff

more progress; initial stab at calling hscMain from the driver

Summary:

  * in normal mode, the driver calls hscMain directly.  The
    compilation manager is not involved at all.  This bit is almost
    there, just waiting for hscMain to catch up.

  * in `ghc --make' mode, the driver will hand off control to the
    compilation manager, and provide the `compile' interface.  This
    bit isn't there yet.

  * in `ghc --interactive' mode, the driver will hand off control
    to the user interface (which doesn't exist yet), which will in
    turn invoke the compilation manager, which in turn invokes the
    driver again through the `compile' interface.  None of this
    is there yet.

Reinstate emptyPCS.

dead code removal

MkDependHS stuff separated into its own file.

Move @compile@-related types to HscTypes.

all compiles now; not quite hooked up to hscMain yet though.

Moving on

Renaming of various previously-TLA'd types.

getting there...

nuke all the search path cruft in this file