We don't want to require that users building source dists have alex/happy

      
      - Ticket #1995: Unsoundness with newtypes
      - Ticket #2475: "Can't unify" error when stopped at an exception
      
      In addition this patch adds the following:
      
      - Unfailingness: RTTI cannot panic anymore. 
        In case of failure, it recovers gracefully by returning the "I know nothing" type
      - A -ddump-rtti flag

This patch adds support for raw token streams, that contain more
information than normal token streams (they contains comments at
least). The "lexTokenStream" function brings this support to the
Lexer module. In addition to that, functions have been added to
the GHC module to make easier to recover of the token stream of 
a module ("getTokenStream").

Building on that, I added what could be called "rich token
stream": token stream to which have been added the source string
corresponding to each token, the function addSourceToToken takes
a StringBuffer and a starting SrcLoc and a token stream and build
this rich token stream. getRichTokenStream is a convenience
function to get a module rich token stream. "showRichTokenStream"
use the SrcLoc information in such a token stream to get a string
similar to the original source (except unsignificant
whitespaces). Thus "putStrLn . showRichTokenStream =<<
getRichTokenStream s mod" should print a valid module source, the
interesting part being to modify the token stream between the get
and the show of course.

Now that the Cabal file is generated by configure, it would be nice
if clean worked even if the cabal file is missing. So now we just rm -rf
the dist directory.

This allows us to put the proper version number into them

This bug has been hanging around for a long time, as you'll see by its
number. The fix implements a feature that is really needed by SYB3, to
allow an instance to (rather indirectly) refer to itself.  The trickiness
comes when solving the superclass constraints.

The whoel issue is explained in Note [Recursive instances and superclases]
in TcSimplify.

In cracking this one I found I could remove the WantSCs argument to the
ReduceMe flag, which is a worthwhile simplification.  Good!

The DPH library tripped an ASSERT.  The code is actually OK, but it's
badly-optimised so I changed it to WARN.  The issue here is explained
in ClosureInfo, Note [Unsafe coerce complications].

I was perplexed about why an arity-related WARN was tripping. It took 
me _day_ (sigh) to find that it was because SimplEnv.substExpr was taking
a short cut when the substitution was empty, thereby not subsituting for
Ids in scope, which must be done (CoreSubst Note [Extending the Subst]).

The fix is a matter of deleting the "optimisation".  Same with
CoreSubst.substSpec, although I don't know if that actually caused a
probem.

We like things with high arity, so when doing eta reduction
it's probably a good idea to avoid reducing arity.

This warning tests that the arity of a function does not decrease.
And that it's at least as great as the strictness signature.

Failing this test isn't a disater, but it's distinctly odd and 
usually indicates that not enough information is getting propagated
around, and hence you may get more simplifier iterations.

If you say -ddump-xx we effectively add -fforce-recomp, so that you
see your dump output.  But this works badly in --make mode, because
you get the ddump output for every module, which is probably not what
you want.  This patch forces recompilation with -ddump-X only in one-shot
mode.  

Of course, this only affects people using -ddump options.

Having 2.0.1 causes some unicode tests to fail

In the ext-core parser I guess I never tested:
* existential type variable bindings in case alts
* empty data declarations

That'll learn me!

It was calling openTempFile which uses a 600 permissions mask.

Rename function accordingly.

The callback wasn't very flexible to begin with.  There's pretty much
no way around to calling 'compile' inside that callback since
'upsweep' depends on certain side effects of compile.  It therefore
makes more sense to restrict the callback to its intended use only,
namely to log warnings and errors.