- debugging: print weights as unsigned
- clean up getStablePtr a bit
- correct a comment

Fix a bug in freeStablePtr, the weight being subtracted from the entry
in the stable pointer table was wrong.

Fix a couple of bugs in the StablePtr weigthed reference counting.

turn off -Wpointer-arith, it generates some warnings in the headers on RH6.2

Convert one of the alternatives in an algebraic switch into the
default, if we don't already have a default.  This generates better
code in particular for inline comparison primops.

Noticed by: Qrczak

Use mkAbsCStmts rather than AbsCStmts directly.

test for bogus parse caused by Happy bug

ForeignObjs were incorrectly passed to foreign functions by the NCG.
Fixed.

Make freeze and thaw top-level

Put the max in a better place

Add comment

Document -ddump-minimal-imports

Add test for selectors in data/newtype with contexts

wrong2 wrongly referred to as wrong1 - righted

Don't use Int# values larger than 2^31.

Fix a bug reported by Jose Emilio Labra Gayo

	newtype Foo a => T = MkT (out :: a)

The selector 'out' was being given an incorrect RHS.
(Core Lint spotted it.)

Move prim_ZONE and prim_GMTOFF into Hugs-only code

Remove redundant imports and dead code

Remove redundant imports

Add filterFB comment

Don't use a typedef called int64 in RtsAPI.
It conflicts with e.g. OCaml's headers.
It should be cleaned better...

Fix for copying html documentation into a binary dist.

Add a test (also benchmark) for threadDelay, Random, and QSemN.  This
test starts a large number of threads which all wait for a random
delay.  The semaphore is used to wait for them all to finish before exiting.

Change the way threadDelay# is implemented.

We now use a list of sleeping threads sorted in increasing order by
the time at which they will wake up.  This avoids us having to
traverse the entire queue on each context switch.

Sorry Sigbj&oslash;rn, but DocBook won't accept &oslash; in <Author>
elements. Seems to work everywhere else...

Don't strip libgmp.a. (But perhaps --strip-unneeded or something
would be harmless.)

PS. The configure script looks for __gmpz_fdiv_qr, but gmp that I have
installed uses the name mpz_fdiv_qr and is thus not being found. (Also
it is dynamically linked, where ghc's gmp is only statically linked.)
I guess that configure should check for either name.

Added more authors.

Let foreign import dynamic accept a newtyped Addr too.

Add two tests for signatures without definitions.

Add a test for Directory.getPermissions.

Change implementation of Directory.getPermissions to use access(2)
rather than stat(2).  This is rather more sensible as the permissions
returned will be relevant to the current user rather than the owner of
the file.

Fix a problem where ^C wasn't recognised while waiting for I/O.

Fix sparc NCG to track recent NCG switch table reg-alloc bug fix.

Add regression check for recent NCG switch table reg-alloc bug fix.

Don't do cg026 on sparc; it doesn't work with the NCG (foreign label stderr).

Wibble.

Make prelude/PrimOp.lhs, not prelude/PrimOp.o, depend on $(PRIMOP_BITS).
'make boot' doesn't work otherwise, since the primop includes don't get
built.

Works now independent of whether GNU make is installed as gmake or make on the
system.

[lewie: It was not so good to just change `make' to `gmake'; broke the thing
on our Solaris boxen, which have GNU make as the default make.]

Make the register allocator deal properly with switch tables.
Previously, it didn't calculate the correct flow edges away from the
indirect jump (in fact it didn't reckon there were any flow edges
leaving it :) which makes a nonsense of the live variable analysis in
the branches.

A jump insn can now optionally be annotated with a list of destination
labels, and if so, the register allocator creates flow edges to all of
them.

Jump tables are now re-enabled.  They remain disabled for 4.08.1,
since we aren't fixing the problem properly on that branch.

I assume this problem wasn't exposed by the old register allocator
because of the live-range-approximation hacks used in it.  Since it
was undocumented, we'll never know.

Sparc builds will now break until I fix them.

look for GMP 3, not GMP 2.