Re: libverbose_terminate, pass 1 of N

[Phil Edwards <phil at jaj dot com>]

On Fri, Dec 20, 2002 at 07:43:40PM -0800, Nathan Myers wrote:
> On Sat, Dec 21, 2002 at 03:26:14AM +0100, Gabriel Dos Reis wrote:
> > Phil Edwards <phil@jaj.com> writes:
> > 
> > | Jason's patch
> > | 
> > |     http://gcc.gnu.org/ml/gcc-patches/2002-12/msg00957.html
> > | 
> > | started an interesting thread of about a dozen messages.  The upshot is that
> > | currently, using __verbose_terminate() as the termination handler requires a
> > | change to user code, usually main().  It Would Be Cool[tm] if the user could
> > | install this handler at link-time with -l, or at runtime with LD_PRELOAD.
> 
> Have we established that a verbose-termination-by-default would not
> be conformant?  Is there some other reason not to want it as the
> default?

Verbose termination calls fprintf, which can't be used in a freestanding
environment.  If -ffreestanding (or whatever it is) were to cause a macro
to be defined, then I see no reason why we couldn't do

    // libsupc++/eh_terminate.cc
    std::terminate_handler __cxxabiv1::__terminate_handler =
    #if freestanding
                                        std::abort;
    #else
                                        __gnu_cxx::__verbose_termination_handler;
    #endif

The only note I can find in the standard is "default terminate() calls
abort," which doesn't seem to preclude also doing other actions.


Phil

-- 
I would therefore like to posit that computing's central challenge, viz. "How
not to make a mess of it," has /not/ been met.
                                                 - Edsger Dijkstra, 1930-2002