[cap-talk] Persistence as a cap value (was: Re: ...PLASH discussion)

[Karp, Alan H]

Sandro Magi wrote:
>
> Were remote machines assumed to be running a trusted runtime? If not,
> how did the GC work? (assuming you can divulge that of course)
>
Of course I can.  All our code was released under GPL or LGPL.  We did not assume a trusted runtime on the other system.  All interaction with a remote user was through a proxy running locally as a normal client.  Since all the entries were local, we could use a local algorithm.  There was still the possibility that the remote user had dropped the reference, but there is never a way to distinguish that case from holding a reference that will never be used.  The proxy's quota was charged for entries in it's c-list, so the remote user had an incentive to remove unnecessary entries.

________________________
Alan Karp
Principal Scientist
Virus Safe Computing Initiative
Hewlett-Packard Laboratories
1501 Page Mill Road
Palo Alto, CA 94304
(650) 857-3967, fax (650) 857-7029
http://www.hpl.hp.com/personal/Alan_Karp