Re: [boost] [smart_ptr] questions about some regression test cases

From: Frank Mori Hess (frank.hess_at_[hidden])
Date: 2008-04-28 12:11:48

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Monday 28 April 2008 09:27 am, Peter Dimov wrote:
> Daniel Frey:
> > Consider:
> >
> > class Object {
> >  static map< string, shared_ptr< Object > > cache;
> >
> >  Object()
> >  {
> >    cache.insert( make_pair( "foo", shared_from_this() ) );
> >    throw 42;
> >  }
> > };
> >
> > With the above BOOST_ASSERT(_shared_count.empty()), ::abort() seems to
> > most sensible thing we can do, otherwise it would lead to very dangerous
> > situations where cache contains a shared_ptr to a deleted object.
>
> The original assert catches this case, does it not?

Yes.  The difference would be if cache was:

static map< string, weak_ptr< Object > > cache;

Then the current assert would allow it (and it would probably be safe at least 
in a single-threaded context), but BOOST_ASSERT(_shared_count.empty()) would 
abort.

- -- 
Frank
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFIFfdY5vihyNWuA4URAvhbAJ9MzZO+arh/mrbr6i0KCRXF6QM4zQCfS/EA
1rReoupBhiB/cStdvJQJeU8=
=RWWF
-----END PGP SIGNATURE-----