Re: Free'ing extended types

[bobduff@world.std.com (Robert A Duff)]

In article <4xpw8qvlau.fsf@lagrange.enst-bretagne.fr>,
Laurent Guerby <Laurent.Guerby@enst-bretagne.fr> wrote:
>   Note that the impact of a  GC on the  "erronenousness" of a program
>is    interesting.  An   Ada run-time  with   GC    can  simply ignore
>Unchecked_Dealocation calls, or do smart thing and raise Program_Error
>when an access to a previously deallocated object is performed. In the
>first case (do  nothing on deallocation)  I think the execution is not
>erroneous because RM95-13.11.2(16) says :
>
>"Evaluating  a  name that denotes a  nonexistent  object is erroneous.
>The execution  of a call to  an  instance of Unchecked_Deallocation is
>erroneous if the object was created other than by  an allocator for an
>access type whose pool is Name'Storage_Pool."
>
>   On  a garbage collected implementation,  the object is likely to be
>"existant" or your GC is wrong ;-).

No, I don't think so.  The RM says an Unchecked_Deallocated object is
non-existent, and deallocating it is erroneous.  If you have a GC that
thinks it still exists, that's fine, but it's irrelevant to the formal
definition of erroneous.  (In this case I'd like the gc to raise an
exception, but of course that's not required, since erroneous means
*anything* can happen, including keeping the officially deallocated
object around and using it as usual.)

- Bob