Re: DISCUSS: MUST reject in OCSPv1

["Florian Oelmaier" <oelmaier@xxxxxxxxxxx>]

> Florian -
> 
> Sorry, my misunderstanding.  I think that your strategy is also good if 
> both the client and server implement it, but it sounds like your clients 
> would have the same backward-compatibility problem if talking to someone 
> else's old servers.  I.e. an attacker could send a nonceless request to 
> an old existing server and record the nonceless response, and then 
> replay that to your client to make you think that that server is 
> securely signalling that it doesn't support nonces.

/agree. Thats completely correct. 

The difference is that our strategy does not require new clients as it does not require anything "new" on the client side. But you are right, if a client talks to an "old" server the attack described is as feasible as with Russī proposal. Only "new" servers can prevent this type of attack in both proposals. My bad, sorry.

-- 
Florian Oelmaier
SyTrust