I'm experimenting with making GHC a bit more aggressive about
  a) dropping case expressions if the result is unused
        Simplify.rebuildCase, CaseElim equation

  b) floating case expressions inwards
        FloatIn.fiExpr, AnnCase

In both cases the new behaviour is gotten with a static (debug)
flag -faggressive-primops.  The extra "aggression" is to allow
discarding and floating in for side-effecting operations.  See
the new, extensive Note [PrimOp can_fail and has_side_effects]
in PrimoOp.

When discarding a case with unused binders, in the lifted-type
case it's definitely ok if the scrutinee terminates; previously
we were checking exprOkForSpeculation, which is significantly
worse.

So I wanted a new function CoreUtils.exprCertainlyTerminates.
In doing this I ended up with a significant refactoring in
CoreUtils.  The new structure has quite a lot of nice sharing:

    exprIsCheap             = exprIsCheap' isHNFApp
    exprIsExpandable        = exprIsCheap' isConLikeApp

    exprIsHNF               = exprIsHNFlike isHNFApp
    exprIsConLike           = exprIsHNFlike isConLikeApp
    exprCertainlyTerminates = exprIsHNFlike isTerminatingApp

There was a trivial typo which meant that important
newly-added rules would never fire!

is used for optimisation. (enabled by default)

Two changes here

* The main change here is to enhance the FloatIn pass so that it can
  float case-bindings inwards.  In particular the case bindings for
  array indexing.

* Also change the code in Simplify, to allow a case on array
  indexing (ie can_fail is true) to be discarded altogether if its
  results are unused.

Lots of new comments in PrimOp about can_fail and has_side_effects

Some refactoring to share the FloatBind data structure between
FloatIn and FloatOut

This patch implements the idea of deferring (most) type errors to
runtime, instead emitting only a warning at compile time.  The
basic idea is very simple:

 * The on-the-fly unifier in TcUnify never fails; instead if it
   gets stuck it emits a constraint.

 * The constraint solver tries to solve the constraints (and is
   entirely unchanged, hooray).

 * The remaining, unsolved constraints (if any) are passed to
   TcErrors.reportUnsolved.  With -fdefer-type-errors, instead of
   emitting an error message, TcErrors emits a warning, AND emits
   a binding for the constraint witness, binding it
   to (error "the error message"), via the new form of evidence
   TcEvidence.EvDelayedError.  So, when the program is run,
   when (and only when) that witness is needed, the program will
   crash with the exact same error message that would have been
   given at compile time.

Simple really.  But, needless to say, the exercise forced me
into some major refactoring.

 * TcErrors is almost entirely rewritten

 * EvVarX and WantedEvVar have gone away entirely

 * ErrUtils is changed a bit:
     * New Severity field in ErrMsg
     * Renamed the type Message to MsgDoc (this change
       touches a lot of files trivially)

 * One minor change is that in the constraint solver we try
   NOT to combine insoluble constraints, like Int~Bool, else
   all such type errors get combined together and result in
   only one error message!

 * I moved some definitions from TcSMonad to TcRnTypes,
   where they seem to belong more

TBAA allows us to specify a type hierachy in metadata with
the property that nodes on different branches don't alias.
This should somewhat improve the optimizations LLVM does that
rely on alias information.

On XCode 4.1, we use /usr/bin/gcc-4.2 as it makes better code than
/usr/bin/gcc for us. However, gcc-4.2 doesn't exist in XCode 4.2, so we
need to use /usr/bin/gcc there. As the installer can be used on either,
we make it always use /usr/bin/gcc.

It was removed in 4.2.

The libraries get configured during "make", so plain old "make clean"
should clean up the configure droppings.

Also fixed another wildcard/patsubst problem.

We were applying $(wildcard ...) to the literal string
libraries/%/*.buildinfo, which of course had no matches.

Patch from Karel Gardas!

We now carry around with CmmJump statements a list of
the STG registers that are live at that jump site.
This is used by the LLVM backend so it can avoid
unnesecarily passing around dead registers, improving
perfromance. This gives us the framework to finally
fix trac #4308.