Collation implementation

Nathan Myers ncm@cantrip.org
Mon Jun 19 11:34:00 GMT 2000


On Mon, Jun 19, 2000 at 08:58:47AM -0700, Ulrich Drepper wrote:
> Dietmar Kuehl <dietmar_kuehl@yahoo.com> writes:
> 
> > I know that Ulrich Drepper has a different idea how to address this
> > issue but I don't share his view, especially in the light that I think
> > the right way is to base the standard C library on the standard C++
> > library and not vice versa
> 
> Get real.  You never ever want to base everything else on a C++
> library.  This is a horrendous stupidity.  C++ brings not only an
> unstable API, it also introduces a huge amount of overhead (code,
> data, initialization).

Dietmar is not stupid, never mind horrendously stupid.

The C++ library is free to provide its own library facilities, and
this "huge amount of overhead" has not been demonstrated.  It's 
probably quite a bit smaller than (e.g.) having two GUI widget sets.  
In any case, there are no standard C library facilities to use to 
implement the required C++ semantics, and glibc with its extensions 
is not universally available.

Certainly it would be best if both C and C++ libraries used the same 
data files underneath, whatever code is used to apply them.  It may be 
that the IBM package would be preferable as the locale implementation 
underlying both C and C++ libraries.

Nathan Myers
ncm at cantrip dot org



More information about the Libstdc++ mailing list