Re: Fw: [patch] Make std::tr1::shared_ptr thread-safe.

["Peter Dimov" <pdimov at mmltd dot net>]

Paolo Carlini wrote:
> Peter Dimov wrote:
>
> > I'll try to review basic_string and locale::facet so that I can speak
> > in less general terms.
>
> Yes, by now I think I understand the issue at large. Of course a
> review of the usages in basic_string and locale::facet is very
> welcome.

> From what I see... it's extremely unlikely that the issue will cause 
problems in practice. I can't come up with a realistic example that has even 
a small chance to "fail".

Constructing such tests with shared_ptr is relatively easier because I can 
vary the object destructor, something like

struct X
{
   int v;

   X(): v( 0 ) {}
   ~X() { v = 1; }
};

and with several billion repetitions of

assert( px->v == 0 );
px.reset();

from two threads simultaneously on an Alpha MP machine it might fail. Even 
this isn't easy to set up, because the two threads need to execute their 
sequences at exactly the same time using two shared_ptr instances sharing 
ownership of the same object.