This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: [PATCH] Move rope, slist, hash_map, hash_set to __gnu_cxx
- From: Paolo Carlini <pcarlini at unitus dot it>
- To: Gabriel Dos Reis <gdr at codesourcery dot com>
- Cc: libstdc++ at gcc dot gnu dot org
- Date: Sun, 30 Dec 2001 16:27:49 +0100
- Subject: Re: [PATCH] Move rope, slist, hash_map, hash_set to __gnu_cxx
- References: <3C2EDC47.B84A1816@unitus.it> <flzo40ud9a.fsf@riz.cmla.ens-cachan.fr>
Gabriel Dos Reis wrote:
> Paolo Carlini <pcarlini@unitus.it> writes:
>
> | Hi,
> |
> | this one moves such estensions (i.e., all those already present in include/ext
> | before the recent addition of algorithm) to namespace __gnu_cxx. Tested
> | i686-pc-linux-gnu (*).
>
> Looks OK.
Thanks. As I'm learning from the lists, right now the compiler does not bootstrap
on x86. It is not good practice committing even completely unrelated library
patches in such circumstances, right?
I'm going therefore to wait for Benjamin's point of view on backward/algo.h, then,
as soon as the compiler boostraps again, I will take action.
> I'm deferring the issue of handling existing codes that assume
> extensions are in std::. I don't know whether whether simple
> documentation is sufficient or if some compiler magic are
> necessary.
I don't know either.
In general, if the legacy code accessed the extensions via using declarations it's
easy to just change them, but if it was written bringing in all std:: via an using
directive or accessing single names by prepending std:: it is probably necessary to
grep for each name...
With Phil we can prepare ASAP documentation for 3.1 describing this.
Cheers,
Paolo.