[cap-talk] Stack walking is capability access control?

[Sandro Magi]

David Wagner wrote:
> Geoffrey Alan Washburn <geoffw at cis.upenn.edu> 
>> 	I never claimed it was equivalent to a capability system.  I only was 
>> claiming that it can be treated as an instance of a capability system. 
>> The text in my proposal, as quoted by Sandro, does unfortunately not 
>> clearly articulate this distinction.
> 
> I don't follow the distinction you are making.  I would say that Java is
> not a capability system, and it is not an instance of a capability system
> (whatever that would mean).  Can you explain?  Maybe we are using the
> same words but meaning something different by them.

In a private communication to me earlier, Geoffrey clarified:

  To further clarify, I am not claiming that stack inspection is
  equivalent to capability based access control, but that it may be
  implemented using capability based access control.

This is theoretically true, in the Turing-equivalence sense of whole
program transformations, as others have mentioned.

Practically speaking however, it does not conform to important security
practices such as POLA, nor does it look like a capability system to the
developer. Geoffrey has said he would clarify his statements in
follow-up papers, so I hope this discussion has highlighted the precise
distinctions to be made.

Sandro