stack-protector guard location

[Roland McGrath]

> I don't see a problem with giving it a name.  We can always define an
> obscure namespace.  For example, __GCC_stack_guard.  I think it's
> reasonable for us to claim __GCC_ as a special subspace of the
> implementation namespace.

I think __gnuc_ has the precedent there.  But sure, I'd certainly agree
that __GCC_, _GCC_*, __gcc_*, etc. are fine prefix choices.  In fact, I
don't object to claiming __stack_guard__ as part of the implementation name
space for C/C++ (speaking as a maintainer of other parts of "the
implementation", consider us coordinated).  Aside from pedantry about the
general case for other languages, my real objections have to do with the
veracity of what the DWARF info says about the occupants of the name space.


Thanks,
Roland