This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: auto_ptr?
- To: egcs at cygnus dot com
- Subject: Re: auto_ptr?
- From: Nathan Myers <ncm at cygnus dot com>
- Date: Wed, 27 May 1998 22:09:27 -0700
- Newsgroups: cygnus.egcs
- Organization: http://www.cantrip.org/
- References: <199805271639.SAA24329@peti.gmd.de>, <199805272158.OAA24568.cygnus.egcs@atrus.synopsys.com>
Joe Buck wrote:
>
> > In the memory include file I read:
> >
> > | // Note: auto_ptr is commented out in this release because the
> > | // details of the interface are still being discussed by the C++
> > | // standardization committee. It will be included once the iterface
> > | // is finalized.
> > |
> > | #if 0
> > | [...]
> >
> > Is this still true? I was under the impression that auto_ptr has been
> > settled by now. At least it is covered in great detail in Stroustrup's
> > latest edition of "The C++ programming language" and quite a few
> > programs I tried to compile want to use it. Just curious.
>
> It's now settled. However, the settlement wasn't final until November
> 1997, so at the time the file was written the comment was accurate.
>
> I am not certain that Stroustrup's book exactly matches the final
> decision (which came out quite different from the Dec. 1996 draft you'll
> find on www.cygnus.com).
The "final" settlement may be found among Stroustrup's errata:
http://www.research.att.com/~bs/3rd_printing5.html
or in the 5th printing of his book.
However, this is not so final as might be thought: it has some serious
usability problems, and egcs doesn't like it. (Moving the member template
auto_ptr_ref out of the class template helps.) There will probably be a
defect report affecting auto_ptr, but it shouldn't affect code which doesn't
mention auto_ptr_ref.
Nathan Myers
ncm@cantrip.org