This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: Overloading vs. string functions, etc.
- To: Mark Mitchell <mark at codesourcery dot com>
- Subject: Re: Overloading vs. string functions, etc.
- From: Benjamin Kosnik <bkoz at redhat dot com>
- Date: Fri, 25 May 2001 01:14:37 -0700 (PDT)
- cc: aj at suse dot de, gdr at codesourcery dot com, libstdc++ at gcc dot gnu dot org, drepper at cygnus dot com
> Definitely. I'm thinking for the sometime-distant future.
...at that time the config scripts can be hacked to switch from c_std to
c, which just pass-through. That's the idea, at least. Some tweaking will
probably be involved but it should be minimal.
Uli and I talked about this a bit ago, but I can't remember the resolution.
Basically it involves changing glibc's
__BEGIN_DECLS
for c++ to be
namespace std
{
extern "C"
{
and similar changes to __END_DECLS
All in all, this is a simpler approach to solving this correctly than the
contortions c_shadow does. The downside is that then the C++ headers
aren't exactly correct either, as there is extra, non-standard gunk in
the C++ headers (or could be, if things aren't done carefully.)
Note that both Sun and Microsoft use this approach, probably others.
I'm starting to lean towards it as the real solution for gnu-linux.
(throw specs are irrelevant to this as Jason's already got that covered.)
-benjamin