The wrapper functions can end up in interface files, and thus are
part of the ABI hash. But uniqs easily change for no good reason
when recompiling, which can lead to an ABI hash needlessly changing.

With the new register allocation scheme, the LLVM back end must always use
liveness information.

There was a long thread on cvs-ghc which concluded that although it's
the Right Thing to deprecate Rank2Types and PolymorphicComponents, in
favour of RankNTypes, it would cause lots of busy-work for library
authors and the pain isn't worth the gain.

So this patch removes the deprecation, and documents the synonym-ity.

This means that -static now disables the -fPIC that we're currently
getting by default.

We agreed that it's not worth the bother of trying to maintain all
these distinct flags; RankNTypes will do the job fine.  Trac #6032.

We were being inconsistent about how we tested whether dump flags
were enabled; in particular, sometimes we also checked the verbosity,
and sometimes we didn't.

This lead to oddities such as "ghc -v4" printing an "Asm code" section
which didn't contain any code, and "-v4" enabled some parts of
"-ddump-deriv" but not others.

Now all the tests use dopt, which also takes the verbosity into account
as appropriate.

Mostly d -> g (matching DynFlag -> GeneralFlag).
Also renamed if* to when*, matching the Haskell if/when names

This avoids confusion due to [DynFlag] and DynFlags being completely
different types.

It was turning on Opt_D_dump_cmmz_cbe rather than Opt_D_dump_cmmz_cfg

We instead link objects into a temporary DLL and dlopen that

I also removed the default values from the "Discounts and thresholds"
note: most of them were no longer up-to-date.

Along the way I added FloatSuffix to the argument parser, analogous to
IntSuffix.

This is an ugly kludge to make a DynFlags value available for the
'trace' functions. It may not be the value we really ought to use,
but it'll be good enough for the pretty-printer to use.

Ideally we'd pass the real DynFlags down to all the trace calls,
but this will do for now at least.

This required various build system changes to get the build to go
through.

In the inplace shell wrappers, we set LD_LIBRARY_PATH to allow programs
to find their libraries. In the future, we might change the inplace tree
to be the same shape as an installed tree instead. However, this would
mean changing the way we do installation, as currently we use cabal's
installation methods to install the libraries, but that only works if
the libraries are under libraries/foo/dist-install/build/..., rather
than in inplace/lib/...

This makes it easier to ensure that we get consistent consistency
checking, e.g. that
    -f1 -f2
will do the same checks as
    -f2 -f1

I think that some of the checks were bogus before, but hopefully
all are correct now.

It's still the case that
    -dynamic -static
will leave -fPIC enabled, but it's tricky to avoid that, and we already
have similar situations where for example
    -XRankNTypes -XNoRankNTypes
will leave ExplicitForAll enabled.

We used to add the options when the way was enabled, but this caused
problems with the static/dynamic choice, as once the 'dynamic' options
had been added it wasn't easy to remove them. We therefore didn't allow
-static on the commandline if we'd already seen -static, which kludged
around the issue.

But apart from being unsatisfactory in the first place, this is a
problem if we want to allow -dynamic being on by default, as there
wouldn't be any way to turn it off.

So now we add C/CPP/linker flags required by the 'way's when flags
are requested, rather than during flag parsing.

The current fix is relatively dumb as far as where to add HpLim
checks: it will always perform a check unless we know that we're
returning from a closure or we are doing a non let-no-escape case
analysis.  The performance impact on the nofib suite looks like this:

            Min          +5.7%     -0.0%     -6.5%     -6.4%    -50.0%
            Max          +6.3%     +5.8%     +5.0%     +5.5%     +0.8%
 Geometric Mean          +6.2%     +0.1%     +0.5%     +0.5%     -0.8%

Overall, the executable bloat is the biggest problem, so we keep the old
omit-yields optimization on by default. Remember that if you need an
interruptibility guarantee, you need to recompile all of your libraries
with -fno-omit-yields.

A better fix would involve only inserting the yields necessary to break
loops; this is left as future work.

Signed-off-by: Edward Z. Yang <ezyang@mit.edu>

This single commit combines a lot of work done by
Thijs Alkemade <thijsalkemade@gmail.com>, plus a slew
of subsequent refactoring by Simon PJ.

The basic idea is
* Add a new expression form "_", a hole, standing for a not-yet-written expression
* Give a useful error message that
   (a) gives the type of the hole
   (b) gives the types of some enclosing value bindings that
       mention the hole

Driven by this goal I did a LOT of refactoring in TcErrors, which in turn
allows us to report enclosing value bindings for other errors, not just
holes.  (Thijs rightly did not attempt this!)

The major data type change is a new form of constraint
  data Ct = ...
    	  | CHoleCan {
    	      cc_ev       :: CtEvidence,
    	      cc_hole_ty  :: TcTauType,
    	      cc_depth    :: SubGoalDepth }

I'm still in two minds about whether this is the best plan. Another
possibility would be to have a predicate type for holes, somthing like
   class Hole a where
     holeValue :: a

It works the way it is, but there are some annoying special cases for
CHoleCan (just grep for "CHoleCan").

This frees wORD_SIZE up to be moved out of HaskellConstants