< Previous by Date Date Index Next by Date >
< Previous in Thread Thread Index  

Re: [reSIProcate] VS2005 crashing with DUM


I'm not willing to buy this without some hard evidence. I have never had a problem with auto_ptr and virtual d'tors, and a Google search turns up nothing. Additionally, I can't think of a plausible reason why this would be the case (auto_ptr<Foo> isn't going to be calling ~Foo() directly).

Best regards,
Byron Campen


auto_ptr is not compatible with virtual destructors. that may be the odd bit.
david

On 10/23/07, Alan Hawrylyshen <alan@xxxxxxxxxxxx> wrote:


On 23-Oct-07, at 04:30 , Nilay Tripathi wrote:
The auto_ptr instance that we create has everything fine, but as soon as we assign it to mClientAuthManager, vfptr goes for a toss. Something creepy in
operator=() of auto_ptr.

Is it safe to assume that the 'creepy' part of auto_ptr is something other
than the 'assignment transfers ownership' semantics of auto_ptr.

Given a class, A, and a pair of auto_ptr's p1 and p2, the following is
roughly true:
 A* p;
 auto_ptr<A> p1(p = new A);
 // p1 'points' to the 'new A' (the same place that 'p' points)
// of course 'p' is dangerous and merely here for pedagogical purposes.
 auto_ptr<A> p2(p1);

 // p2 now 'points' at the same place as 'p' and
 // p1 is pointing at nothing.

auto_ptr<A> p3 = p2 ; // usually disallowed, but has same effect, only p3
 // refers to the same A that p does.

Using auto_ptr<>::release() and auto_ptr<>::get() can put you in trouble
too...

Or was it something else really 'odd'?




Alan Hawrylyshen
a l a n a t p o l y p h a s e d o t c a


_______________________________________________
resiprocate-devel mailing list
resiprocate-devel@xxxxxxxxxxxxxxx
https://list.resiprocate.org/mailman/listinfo/resiprocate-devel

_______________________________________________
resiprocate-devel mailing list
resiprocate-devel@xxxxxxxxxxxxxxx
https://list.resiprocate.org/mailman/listinfo/resiprocate-devel

Attachment: smime.p7s
Description: S/MIME cryptographic signature