This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: Re: New const char* ctor for C++-0x
- From: 3dw4rd at verizon dot net
- To: paolo dot carlini at oracle dot com
- Cc: jwakely dot gcc at gmail dot com, libstdc++ at gcc dot gnu dot org
- Date: Thu, 24 Dec 2009 13:22:00 -0600 (CST)
- Subject: Re: Re: New const char* ctor for C++-0x
Dec 24, 2009 12:40:54 PM, paolo.carlini@oracle.com wrote:
On 12/24/2009 06:19 PM, Ed Smith-Rowland wrote:
> Here is a new patch incorporating the test changes suggested by
> Jonathan and removing the stupid paste-o size_t thing.
Ok, thanks, I think it can go in.
> Regtested and bootstrapped of course.
>
> In addition, for discussion, is a patch with the umsigned long long
> changes. I made the ctor available ic C++-x *and* have long long. So
> in we're doing C++-0x and long long is not available you should still
> get the unsigned long ctor.
In general terms, frankly, this doesn't make sense: C++0x, among tons of
different things, also standardizes long long, thus it cannot be
unavailable in C++0x mode.
Paolo.
---------------
So, a target can't have C++-0x if _GLIBCXX_USE_LONG_LONG is undefined?
So _GLIBCXX_USE_LONG_LONG must be defined if __GXX_EXPERIMENTAL_CXX0X__ is defined?
Or do I not understand the defines?
If one implies the other then sure, I'll skip the LONG_LONG check.
Ed