Re: [critical libstdc++ regression]: Duplicate definitions with -On, n > 1

[Gabriel Dos Reis <gdr at codesourcery dot com>]

Andreas Jaeger <aj@suse.de> writes:

| Gabriel Dos Reis <gdr@codesourcery.com> writes:
| 
| > Gabriel Dos Reis <gdr@codesourcery.com> writes:
| > 
| > | Both GCC-3.0 and GCC-3.1 fail to translate the program below when
| > 
| > Fixed with this.
| 
| Thanks a lot for looking into this.  Is it really correct that glibc
| should not define those macros?  In that case we could change glibc's
| inline functions also.  Or is it this just a workaround for an even
| bigger problem?

It is a workaroung for a bigger problem -- we have serious problems on
building acceptable C++ headers on top of target C headers.  Given, the
timetable for GCC-3.0 release, I thought it was an acceptable
short-time solution.

The problem isn't that glibc defined that macro.  The problem is that
glibc defines something in the global namespace where V3 wants to
have it defined in a particular namespace. 

| Can you tell me exactly which inline functions glibc is allowed to
| declare in C++?  glibc has also a number of string inline functions.

As you asked on the V3 list, this problem is just another instance
of the problem being discussed on the V3 list.  I think we should come
up with a unified approach.  Let's see the outcome of the discussion
on V3 list.

Thanks,

-- Gaby
CodeSourcery, LLC                       http://www.codesourcery.com