From: nee Spangenberg (dsp_at_[hidden])
Date: 2004-01-22 05:18:39
Beman Dawes schrieb:
> At 10:41 AM 1/21/2004, Peter Dimov wrote:
>
> >Yes, shared_array can be extended that way. But my point was much
> simpler:
> >you can assert(i >= 0) in operator[](ptrdiff_t i) to catch s[-1], but you
> >can't if i is size_t, and s[-1] still works. :-)
>
> IIRC, that was the rationale given by Bjarne and others for the desire to
> use of a signed type in similar cases. He was complaining about some
> library code that used size_t as an array index, precluding a simple
> assert(i >= 0).
Just jumping in, because this part of the thread remembers me to the
non-compliant signature of operator[] in boost::auto_array, which I
also mentioned in another thread. Has that issue been fixed? (Here the
same arguments for ptrdiff_t apply as for other smart_array classes, which
are not aware of the actual size of the container it hosts)
Greetings,
Daniel