This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: New STL implementation from SGI
- To: egcs at cygnus dot com, kunert at physik dot tu-dresden dot de
- Subject: Re: New STL implementation from SGI
- From: mrs at wrs dot com (Mike Stump)
- Date: Mon, 13 Jul 1998 17:09:13 -0700
> Date: Mon, 13 Jul 1998 16:32:55 +0200
> From: Thomas Kunert <kunert@physik.tu-dresden.de>
> To: egcs@cygnus.com
> I hope this doesn't sound too stupid:
> What is the purpose of reference-counted strings? For portable code,
> one cannot depend on fast copy-operations
This is called a ``quality of implementation'' issue. Sure, people
cannot rely safely of such things, but some people just don't care a
whole lot about performance and they are happy to give money to the
compiler company that gives them the best performace for their
application.
Or look at it a different way, sure you can portably rely upon
reference counted strings, because if a C++ compiler doesn't offer it,
you can just use g++, as _it_ is portable (for some definition of
portable).
> and if one writes code especially for egcs, rope is better
> anyway.
Ah, but that class is nonstandard (so far). We might add it next time.