Re: [v3] C99 complex math vs. std::complex

[Benjamin Kosnik <bkoz at redhat dot com>]

> I think it's safe to remove it because Benjamin removed back in 2001 the
> relevant autoconf tests as obsolete: therefore nothing currently defines the
> macro at install time and, by the way, it's also completely undocumented.

Agreed. Sorry, I should have been more careful when I removed this in 2001.

Can you bring this part of the patch over to the 3.4 branch as well please?

> Anyway, the below is the complete set of tweaks that I have regtested on
> x86-linux (ongoing x86-64 too).

I too wondered about adding the complex bits to ac_enable_c99. However,
I think this is actually a good idea, like Paolo.

I was concerned that this might impact what systems have wchar_t support
turned on, but it turns out that is separate. Or, kinda separate due to
twisted config bits, but not likely to change with this patch. Maybe we
should just switch to _GLIBCXX_USE_C99_foo where foo is
(math|wchar_t|complex) instead of what we have now (which is mostly
this). Anyway: that's for another day.

So, go for it. While you're changing acinclude.m4, can you change my
comment 

  # Check for the existence of <math.h> complex functions.

to

  # Check for the existence of <complex.h> complex functions.

thanks.

-benjamin