Re: Discuss comments on draft-ietf-pkix-logotypes

[Russ Housley <housley@vigilsec.com>]

Ted:

You make a good argument for additional information in RFC 3280.  I think 
that RFC 3280 ought to be updated to provide better information on 
UTF8.  Perhaps we can ask the PKIX working group to develop an updated 
version of RFC 3280 to handle both topics.  More will undoubtedly surface 
one the document is "open."

The issue is bigger than the logotypes document, and I would not like to 
see it delayed over this more general issue.

Russ

At 10:23 AM 9/19/2003 -0700, hardie@qualcomm.com wrote:
> At 5:19 PM -0400 09/18/2003, Russ Housley wrote:
> > Margaret:
> >
> > RFC 3280 is silent on the presentation of information from certificates 
> > that fail to validate.  This is partly because there are so many 
> > different reasons that a certificate might fail to validate, and it is 
> > partly because it is not a "bits on the wire" issue.
> >
> > I am opposed to the words that Steve proposed because certificate 
> > information is "used" to generate error messages.  Let's face it, most 
> > users do not know what a certificate is.  Anything that helps them 
> > understand that the stuff that they got to enable the use of the 
> > Starbucks wireless network is not working any more, probably because the 
> > certificate expired, is helpful.
> >
> > Russ
>
> Russ,
>         I personally believe that there is a difference between the 
> information
> in the certificates being used to present clear error information and 
> being used
> to present persuasive material.  I think presenting information saying 
> "These folks
> asserted that they are Starbucks, but the certificate is expired, so the best
> I can say is that they were Starbucks at some point in the past" is useful
> and allows an application to take something from the certificate and say
> something sensible about it.
>         Saying "These folks say they are Starbucks, but I can't confirm it;
> here's what else they say:  they're based in Seattle, they have this logo, and
> they trade under this stock exchange symbol".  All of the rest of the data
> is public information, easily attainable, but not trusted here  and not useful
> for the purposes of identifying who these folks are.  Presenting it could 
> easily
> give the end user a false sense of trust--"Ah, yes, I recognize all those 
> attributes
> of Starbucks, so I'll go ahead and click yes".
>         If 3280 is silent about presentation of information from a cert that
> fails to validate because there may be multiple reasons, I can see your
> concern about introducing it here.  But it strikes me that the right thing
> to do is to work through that problem for the general case.  I think this
> particular example makes a strong case for why.
>                         best regards,
>                                         Ted Hardie