In message <20050731101751.GC24014@drowsy.duskware.de>, Martin Husemann writes:
>On Sun, Jul 31, 2005 at 05:08:40AM -0500, Peter Seebach wrote:
>> It's not the NULL part that's at issue; it's that some processors trap if you
>> load an invalid address into an address register, and invalidate the addresses
>> of freed memory.  So it generates a processor trap, and the question of what
>> would happen if you got to the compare instruction is rendered irrelevant.
>
>I'm not arguing the processor trap part, but I don't believe any 
>implementation would be allowed to risk such a user visible trap on the
>comparison of a pointer against NULL.
>
>Is there any concrete implementation that you think about?

I don't remember the names, but it came up in committee, and the current
wording on "trap representations" was carefully constructed to allow for
this.

The problem is that, on such a system, there's really no way to find out
whether a chunk of data will be valid when loaded as a pointer.  There's
nothing the implementation can do to check the pointer for safety without
loading it...

-s