[cap-talk] Why is EQ so dang fascinating?
tribble at e-dean.com
Fri Nov 2 23:47:03 EDT 2007
On 11/2/07, Charles Landau <clandau at macslab.com> wrote:
> At 8:00 AM -0400 10/27/07, Mark Miller wrote:
> >Although above we show a trivial implementation of join in terms of
> >EQ, Dean has previously posted techniques which can successfully
> >implement join in a system without EQ. ... Since join is adequate
> >for Grant Matching, Dean has indeed successfully answered this
> >longest-standing challenge to his position on the EQ debate.
> What hasn't been shown is a technique that can implement a join
> operation without EQ that completes in a bounded time regardless of
> the capabilities being joined.
Ah. But note that none of the prior scenarios for join required
bounded time, because messages to untrusted participants were already
required in the protocol (e.g., GrantMatcher). In typical cases, the
operation completes in bounded time for cooperating participants, and
only takes unbounded time for malicious participants. With
non-blocking send, that need pose no burden on the evaluator, but the
issues of synchronous send and/or single-use capability remains a
potential problem for some scenarios.
More information about the cap-talk