next snapshot update

[Benjamin Kosnik <bkoz at cygnus dot com>]

I'd like to do a snapshot in approximately two weeks. Here's what I'd like
to have by then:

- MT safe string
- testsuite parity on [x86 powerpc alpha]-linux/cygwin/solaris
- ctype support in for solaris2.5.1 (irix6.5? aix4.2? palmOS?)
- --enable-namespaces on by default

This is a deceptively large ammount of work. I don't know what is
going on with the MT safe string work (Nathan? Phil? Ryszard?) but
would appreciate an update.

Gaby, am I missing anything? Is there anything else that should be
attempted? Is some of the above not a great idea?

I've been working on solaris2.7 testsuite parity, meaning, no
additional fails on solaris2.7 versus the x86/linux baseline. This is
ongoing.

It looks like additional configuration hackery is necessary for
complex<float> to work correctly on alpha. Or, more correctly, not
work but not kill the compiler. Complex support seems pretty abysmal
at the moment for targets that are not x86. In addition, some of the
current hacks (ie, not compiling the body of operator>>(basic_istream
if gcc/g++ would give an ICE) need a more elegant solution, or at
least detailed bug reports, or something. It would be cool if somebody
who had access to an alpha box wanted to hack on this.

Documentation for building on cygwin and cross compilers would be
nice. I'll try to come up with something later this week.

I suspect the rebuilding of libgcc.a will be easier now that libstc++
is in the gcc hierarchy. Can this be simplified?

Anyway. If any major bugs are still outstanding, please let me
know. (I know about the solaris problems, and am working on
them. There are two lingering sss io bugs that I'd like to fix.)

-Benjamin