draft-ietf-sidr-res-certs is pretty clear about requiring the validator to check for certificate revocation, which implies that the validator cannot proceed without the relevant CRL.
As Tim points out, the effective definition of an object being in the collection of a certificate's is that it is present in the manifest; the only other way of defining the product collection even with rsync is to trawl the directory, which may find other objects, including objects deliberately omitted from the manifest; with an alternative protocol such as HTTP (if people are even still thinking about that) there is no standard way of trawling the directory at all. So I do not think it makes sense to make the presence of either CRL or manifest optional in any sense here. Allowing stale CRL and manifest, sure. My validator already allows that, in fact that's the default setting, you have to add explicit configuration to get it to reject anything just because of stale CRLs or manifests. As I think I understand it, this is consistent with the advice we've gotten for years from the PKI experts. _______________________________________________ sidr mailing list [email protected] https://www.ietf.org/mailman/listinfo/sidr
