Re: [IETF-PKIX] Multiple certificates for same key?

[Mark Shuttleworth <marks@xxxxxxxxxx>]

Hiya

I think that saying "OK, but only rarely" is dangerous, because it leads
to implementations which break when they see a signle key in multiple
certs.

How about this:

  1. Most "signing" technologies like S/MIME and PKCS#7 include the
     signing certificate along with the signed data.  So parsing
     applications only have to validate the presented or attached certs.

  2. If the signed package does not include the certs, or references to
     the certs (like the issuer+serial form of authorityKeyIdentifier)
     then the validating technology should examine each of the options
     to see if an appropriate chain of authority can be found before
     crying foul.  This is really the case where the reference is of the
     "hash of key" form, which could in theory point to multiple certs
     all containning the same key.

What I'm trying to say is, we should come up with rules that the software
should follow that allow for multiple certs containing the same key,
rather than saying "hardly ever do this".  The market will determine how
often we do it, and the implementation should support the market, rather
than dictate.  Whew, better get off this pedestal before people start
throwing fruit ;)

--
Mark Shuttleworth
Thawte Consulting