[v3] c++0x includes

[Benjamin Kosnik]

> I didn't tell you already, but the final shape of this looks very nice 
> to me. I will add on top a couple of specific improvement to avoid 
> including too much when -std=c++0x is in effect.

Thanks for the feedback. It would be great if you could fine-tune: I 
think Howard also had some complex.h fixups.

I felt like the response to the RFC was pretty positive, and there'd 
been enough time to get the various other header tweaks and testsuite 
moves shaken out, so this was a good time.... kinda inbetween my travel 
schedule as well. It was also pretty big. Of the initial feedback, I 
think everything is covered except for the non-inclusion of tr1/array 
and array simultaneously: I think I should write some docs on that, no?

What to do about backwards headers is still in-progress. I think we can 
wait a bit more for feedback and clarification, and then figure out what 
to do.

I'll be taking a short break to scope the Alhambra and then I'll be back 
for round 2 (3? 4?) of this to try for the "C" includes in std. So, I'll 
be out of your way for the next week at least... with luck maybe the 
variadic stuff will be in and also TR1 math by the time I get back.

As is, this seems pretty stable, I think. So, I think we are in ok shape.

> Unfortunately I'm seeing some distracting unexpected failures on 
> sparc-sun-solaris2.10:
> 
>    http://gcc.gnu.org/ml/gcc-testresults/2007-03/msg00245.html
> 
> Maybe Kaveh (or Eric) can provide some additional info from the 
> ChangeLog about that (I'm going to wait a bit before committing my 
> already posted patch to not include cstdlib, and will also delay other 
> improvements in queue...)

Yes. I saw those: It looks like AIX has one fail as well.

There appear to be unrelated issues WRT hppa2.0w-hp-hpux11.11 as well: 
maybe this is some fallout from the C pruning?

-benjamin