- From: <tgindin@us.ibm.com>
- Date: Tue, 31 Aug 1999 10:42:11 -0400
- To: "Peter Lipp" <Peter.Lipp@iaik.at>
- cc: "Solo, David" <david.solo@citicorp.com>, "'w3c-ietf-xmldsig'" <w3c-ietf-xmldsig@w3.org>
"Peter Lipp" <Peter.Lipp@iaik.at> on 08/30/99 06:41:43 AM To: "Solo, David" <david.solo@citicorp.com>, "'w3c-ietf-xmldsig'" <w3c-ietf-xmldsig@w3.org> cc: Subject: AW: syntax and processing draft - update Generally, I like the proposal. Just some minor remarks: a) why is the objecttype "hidden" in objectlocation? I don't have a strong opinion here but it would seem cleaner to have it as a sepearate element of signedobjectdata. Even Digestalg and value are sepearated...... however b) Wouldn't it be better to put digestalg and value into one element? What if we want to allow more than one digest for multiple signature algorithms? I would like to allow more than one digest (using different algs) in a signature to protect against broken digest algs without the need to have separate signatures. [Tom Gindin] How about the "composite hash algorithm" idea? There should shortly be some published OID's for such combinations as SHA-1 followed by MD2, and some could be added to mix RIPE-MD with either SHA-1 or one of the MD's. Breaking one of those digest algorithms just reduces the digest strength to that of the other digest algorithm(s), and finding collisions in one (which seems to be the most common kind of break) has little effect unless you can find coincident collisions in the other. For longer signature keys, you can even have three hashes, although there might be some reason not to extend the aggregate hash length beyond half the key length. Have a nice F2F! Peter
Received on Tuesday, 31 August 1999 10:44:56 UTC