On Thu, Jul 01, 1999 at 09:27:28AM -0500, "William H. Geiger III"
<whgiii(_at_)openpgp(_dot_)net> wrote:
Unless all keyservers follow exactly the same policies and these policies
do not depend on the order in which the PGP packets are received, the
number of unresolved or unresolvable differences will increase. And I am
not really sure which policy is the "right" one.
I was not aware that this was going on but it is something that, IMHO,
*must* be addressed. This issue not only affects the key servers but also
the users keyrings. I don't like the idea of the servers making these
types of determinations if they are not doing any verification of the
signatures. It opens up a DOS attack on a key by replacing the self
signature with an invalid signature of a newer date. To be honest I don't
like the idea of the public servers removing anything from an existing
key. It opens up problems that the servers are not in a position to
address.
Both strategies, either keeping the first (revocation) signature or
updating them, is open to DoS. I would prefer the keyservers to keep
(and return) all duplicate (revocation) signatures, but only display
one of them. Then the user's OpenPGP implementation should deal with
it. Duplicate signatures shouldn't break anything, but I would like
to know whether anyone assumes that duplicate revocations might break
something?
-Marcel