On Thu, 16 Feb 2012 13:03, bmoel...@acm.org said: > Oh, in this case it's a self-signature. Werner, the problem (aka feature) > is that expiry according to self-signatures isn't carried forward into > third-party certification signatures -- so if an attacker gets hold of the
That depends on how the third party does the key-signing. OpenPGP allows to provide an expiration date for the third party certification (aka key signing). This solves the problem of OpenPGP "CAs" - it does not solve the general problem of CAs at all. The commonly used WoT semantics don't require you to check the expiration date of a passport or driver license either. The signature expiration dates, as used by some folks, try to add some extra value into their key signatures for no good reason: Either the identity has been verified or not - the identity will not change after the expiration date. Even if you change your name later, back at the key signing time you were known under the certified name. > necessarily cover the expiry date, and unlike X.509 where certifications > always come with *some* notAfter date.) A better name for notAfter would be payableBefore. Shalom-Salam, Werner -- Die Gedanken sind frei. Ausnahmen regelt ein Bundesgesetz. _______________________________________________ cryptography mailing list cryptography@randombit.net http://lists.randombit.net/mailman/listinfo/cryptography