This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: PR3042
- To: Mark Mitchell <mark at codesourcery dot com>
- Subject: Re: PR3042
- From: David Edelsohn <dje at watson dot ibm dot com>
- Date: Sun, 10 Jun 2001 17:58:24 -0400
- cc: Gabriel dot Dos-Reis at cmla dot ens-cachan dot fr, bkoz at nabi dot net, jason at cygnus dot com, libstdc++ at gcc dot gnu dot org
I am testing out my patch right now. I strenuously object to your
proposed patch and I never want to see it committed to any GCC development
sources based on the justification that you have provided so far.
Beyond that, you and Gaby do not seem to want to step back and
even try to understand my arguments in the general sense. For instance, I
don't care about xlC name mangling and that is a red herring argument that
ignores the discussion. There is more to compatibility than whether one
can link object files compiled by the two compilers together. Your
failing to look beyond that one aspect ignores the entire argument.
As far as I have been able to ascertain, xlC (without special
template processing options) currently uses the same semantics for
instantiation of symbols that G++ does. The user experience is the same.
Developers already know that there are inconsistencies in C++
instantiation and/or special compiler options that one must invoke. The
point is not whether one can link the object files together, but whether
the compilers behave in similar ways.
I am trying to follow a middle path of both avoiding turning GCC
for AIX into solely a GNU/Linux compatibility environment and avoiding
making GCC for AIX even more different than other targets.
David