[e-lang] Microsoft's laws of identity
andy.dwelly at safedataco.com
Wed Jul 27 13:30:26 EDT 2005
Mark Miller wrote:
> I agree that you should avoid #1.
> Why do you think you want #4? What do you think it means? Are you
> assuming that the "Communicating Conspirators" problem can be solved?
> #2 and #3 don't need to be built into the capabilities themselves.
> They can instead be provided by the intermediate forwarding objects
> that these capabilities designate. See the "Caretaker" pattern in
> Paradigm Regained.
Actually, I wasn't aware that it was considered an insoluable problem -
the links were enlightening. I believe I follow the argument. I'll have
to discuss the point with the rest of the design team because there is
potentially a legal implication in 'who does the work' rather than 'what
work was done'. The conspiring conspiritor example makes the second
party do the work, even though the work is done at the behest or control
of the third party who was supposed to be excluded. Its a distinction
that makes little sense in algorithmic terms but may have an effect on
Elib is certainly a possibility, however.
More information about the e-lang