Re: [orbitcpp-list] A possible bug?

[Phil D. <ph...@us...>]

Hi Jiva,

In fact this code shouldn't work anyway, since you are using an _var
type to hold the orb, which is a smartpointer and will release the old
orb ref when reassigned to. This means that in your example the orb
ref is released twice.

The problem is that this code still segfaults if a CORBA::ORB_ptr is
used, and that is most likely an orbit bug somewhere. I'll check this
out.

Cheers,

Phil

Jiva DeVoe writes:
 > The following code causes a segfault on the second one.  Why?
 > 
 > CORBA::ORB_var orb; 
 >   cout << "1" << endl; 
 >   orb = CORBA::ORB_init(argc, argv, "orbit-local-orb"); 
 >   CORBA::release(orb); 
 >   cout << "1" << endl; 
 >   orb = CORBA::ORB_init(argc, argv, "orbit-local-orb"); 
 >   CORBA::release(orb); 
 >   cout << "1" << endl; 
 >   orb = CORBA::ORB_init(argc, argv, "orbit-local-orb"); 
 >   CORBA::release(orb); 
 >   cout << "1" << endl; 
 >   orb = CORBA::ORB_init(argc, argv, "orbit-local-orb"); 
 >   CORBA::release(orb); 
 >   cout << "1" << endl; 
 >   orb = CORBA::ORB_init(argc, argv, "orbit-local-orb"); 
 >   CORBA::release(orb); 
 > 
 > Some folks wouldn't do this, but we have unit tests that need to setup
 > and tear down a corba connection for serveral consecutive tests.  This
 > crashes my code.
 > -- 
 > 
 > _______________________________________________
 > orbitcpp-list mailing list
 > orb...@li...
 > http://lists.sourceforge.net/lists/listinfo/orbitcpp-list