Re: Object Capability model for IoT
David Barbour <[email protected]>
| Newsgroups | gmane.comp.capabilities.general |
|---|---|
| Message-ID | <CAAOQMSszaOpHLSkwZFKkiRKOoFgMrRFV69Bo6XteOxXramaF9w@mail.gmail.com> |
On Mon, Jun 23, 2014 at 12:27 PM, Karp, Alan H <[email protected]> wrote: > > Maybe I don’t understand your use case, but I don’t see why you can’t > (lazily) clean up the table to remove entries for dropped objects. > The process of "lazily cleaning up the table to remove entries for dropped objects" is essentially a description of "garbage collection". In distributed systems, where capabilities are communicated between machines, deciding whether an object has been 'dropped' can be a significant challenge - due to latency, messages in transit, network partitioning, security concerns about leaks or covert channels through the GC, etc.. Formally, precise distributed GC undecidable. _______________________________________________ cap-talk mailing list [email protected] http://www.eros-os.org/mailman/listinfo/cap-talk