cygwin-specific path mangling of CWD no longer reqd

The test suite wins again: re-enable empty contexts in data
declarations.

posix/DLPrim now uses configure to check for RTLD_LOCAL.

Sigh, recompilation checking in --make mdoe has been badly broken for
well, probably ever.

The problem is that in GHCi mode there's a restriction on the use of
object files, namely that an object file may only depend on other
up-to-date object files in order to be up-to-date itself, because we
don't have support for linking object code to interpreted code.
Unfortunately we were applying this restriction to --make compilations
too, with the result that far too much recompilation was happening.

MERGE TO STABLE

Update the comment about how to add a new primop.

More validity checking, esp for existential ctxt on data cons

Comments and tracing only

TEXT_BEFORE_HEAP & cygwin: same as for mingw

initDefaultHandlers(): siginterrupt() not supported w/ cygwin

re-add way 'p' to GhcLibWays

mingw32_TARGET_OS -> mingw32_HOST_OS

the platform-specific features to use to compile the compiler sources are dependent on the host, not the target platform

only test for -mno-cygwin if the target is mingw32

Fix readHoleResult bug

Add dlopen() and friends to package 'posix'.
A configure-test for RTLD_NEXT is required. Record if -ldl is
required in posix.conf.

Make the closure types consecutive

A couple of cleanups to the previous change: we should test
TABLES_NEXT_TO_CODE rather than USE_MINIINTERPRETER to enable the
MacOSX "plan C", and use structure field selection rather than array
indexing to get the entry code ptr from the info table.

Changes to support powerpc-apple-darwin

From: Wolfgang Thaller <wolfgang.thaller@gmx.net>

Changes to support powerpc-apple-macosx

From: Wolfgang Thaller <wolfgang.thaller@gmx.net>

Changes to support powerpc-apple-macosx

From: Wolfgang Thaller <wolfgang.thaller@gmx.net>

Changes to support powerpc-apple-macosx.

From Wolfgang Thaller <wolfgang.thaller@gmx.net>.

hc-file-bundle: removed dangling ParseIface.hs pointer

-------------------------------
	Fix bugs in rank-N polymorphism
	-------------------------------

Discussion with Mark showed up some bugs in the rank-N
polymorphism stuff, especally concerning the treatment of
'hole' type variables.

See especially TcMType:
	newHoleTyVar
	readHoleResult
	zapToType

Also the treatment of conditionals and case branches
is done right now, using zapToType

imports

Comments

RTS_MINGW_ONLY_SYMBOLS: added rewinddir()

not using -lpthread w/ mingw

Implement Plan C, with correct code to detect the data and text
sections for MacOS X.
Also add a sanity check in initStorage, to make sure we are able to
make the distinction between closures and infotables.

Restore lost imports for catchJust, ioErrors, accidentally removed
by Sigbjorn's commit. (Hard to spot -- it's only needed for 4.08 and earlier.)

Update a comment only.

findPackageMod: for source imports, restrict ourselves to looking for hi-boots only

import tidyup

maybeHomeModule: if chasing after a source import, _only_ look for hi-boots

Put module dependency info in the commentary

Unravel module loop again (tested this time; sorry)

Unravel accidental module loop (sorry)

Fix 64-bit shift operations.

- Move the declarations of the 64-bit "primops" from PrimOps.h to
  HsBase.h where they more properly belong.

- change the names of the 64-bit shift ops to include the "unchecked"
  prefix

- add checked versions of these primops to GHC.Int and GHC.Word, and
  use them.

- update the FFI declarations in GHC.Int and GHC.Word while I'm there.

More on linear implicit params

Tidier printing routines for Rules

Remove spurious ? in linear implicit parameter printing