In the past I've seen this in an interface file

   foo = error "urk"

but *without* a bottoming strictness info on 'foo'. This WARN just 
checks (non-fatally) for the bad case, so that we can track it down easily

The patch adds this rule:

  seq (x `cast` co) y = seq x y

This is subject to the usual treatment of seq rules. It also makes them
match more often: it will rewrite

  seq (f x `cast` co) y = seq (f x) y

and allow a seq rule for f to match.

In this expression:

  let x = f (g e1) in e2

the simplifier will inline f if it thinks that (g e1) is an interesting
argument. Until now, this was essentially the case if g was a data constructor
- we'd inline f in the hope that it will inspect and hence eliminate the
constructor application. This patch extends this mechanism to CONLIKE
functions. We consider (g e1) interesting if g is CONLIKE and inline f in the
hope that this will allow rewrite rules to match.

Not everything is fixed yet; see #3642

That is, you have to build the library/program without -dynamic first,
to get plain object files, and then build it again with -dynamic.

I still need to check whether any changes to Cabal are required to
make this work.

To get ticky profiling you still have to compile with -ticky (for
those modules that you want to profile), but you can link with either
-debug or -ticky.

GHC.Conc.ensureIOManagerIsRunning now creates an IO manager thread if
one does not exist or has died/exited.

Unfortunately this exposed a problem caused by the fact that we have
two base packages, and hence two IO managers, in GHCi: see NOTE
[io-manager-ghci] in rts/Linker.c.  The workaround can go away if/when
we switch to a dynamically linked GHCi.

I discovered that single-stepping over getArgs gave the wrong answer
sometimes, because we were forgetting to set the program's argv and
progname when resuming at a breakpoint.

Test is ghci.debugger/scripts/getargs.script.

There is now less makefile code duplication.
Also, stage2 packages are registered after GHC, which is important as
they might depend on the ghc package (as dph now does).

Do not merge to 6.12

Merge to 6.12

Merge to 6.12 along with the main DoRec patch

MkCore.mkCoreTupTy moves to TysWiredIn, where it is called mkBoxedTupleTy

This variable is tested when deciding whether or not to add debug to
the list of RTS ways, so it needs to be correctly defined.

This is to do with SPECIALISE pragmas in instance declarations,
which I need to think more about