This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: A completely different approach to EH runtime


On Wed, Feb 21, 2001 at 10:39:36PM -0300, Alexandre Oliva wrote:
> On Feb 21, 2001, "Zack Weinberg" <zackw@Stanford.EDU> wrote:
> 
> > If we made it global, we'd have to rename it into the implementation
> > namespace - __frame_objects or something like that.  Can we safely
> > rename it while maintaining backward compatibility?
> 
> Since they used to be static, I suppose so.

I have my i386-linux system with glibc 2.2 which deliberately exports
__register_frame_info so no one else will have a copy.  That's the rfi
from gcc 2.95.x, it uses the static variable.  Now I build a C++
executable with gcc 3.0's libgcc.a, it wants to use __frame_objects.
Oh, and I throw exceptions from a qsort callback.  Does it work?

Thinking about it, it's probably no worse than the existing problem.

zw


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]