On Wed, 15 Jun 2016 15:56:59 +0100
Eran Messeri <[email protected]> wrote:

> On Wed, Jun 15, 2016 at 4:31 AM, Andrew Ayer <[email protected]>
> wrote:
> 
> > Section 6.6 (get-all-by-hash): It's not clear whether the inclusion
> > proof is to the requested STH or to the returned STH.  I'm pretty
> > sure it's supposed to be the returned STH, but this wording
> > suggests it's the requested STH:
> >
> >         tree_size:  The tree_size of the tree on which to base the
> >         proofs, in decimal
> >
> > In this wording, it's unclear what "selected STH" refers to:
> >
> >         inclusion:  A base64 encoded "TransItem" of type
> >         "inclusion_proof_v2" whose "inclusion_path" array of Merkle
> >         Tree nodes proves the inclusion of the chosen certificate in
> >         the selected STH.
> >
> I've changed the description to say 'returned STH' rather than
> 'selected STH' for more clarity. I didn't change the tree_size
> description as it seems obvious to me this is what the client
> requests, but text to clarify that this may not be what the client
> gets is welcome.

I'll think about it.  Changing "Selected STH" to "Returned STH" is
probably good enough.

> > Sections 10.2.5 - 10.2.7 (compliance): It's unclear what role the
> > TLS Feature extension plays.
> >
> 
> > 10.2.5 says that compliance means "accompanied by at least one valid
> > SCT."
> >
> > 10.2.6 says that if transparency_info is present in the TLS
> > Feature extension, then "CT compliance ... is required" and that
> > "TLS clients MUST treat certificates which fail this requirement as
> > non-compliant."  In other words, non-compliance must be treated as
> > non-compliant, which isn't saying anything.
> >
> > 10.2.7 says that if the certificate chain is non-compliant, the "TLS
> > client MUST refuse the connection," which renders the TLS feature
> > extension unnecessary.
> >
> > I assume the intent is that if transparency_info is in the TLS
> > Feature extension, then TLS clients MUST refuse connections using
> > non-compliant chains.  Otherwise, they MAY refuse connections using
> > non-compliant chains, depending on the client's policy.
> >
> What changes do you suggest here? That section 10.2.6 explicitly
> state that a TLS client must refuse the connection in case of
> non-compliance?

I've opened a pull request on GitHub for my suggested changes:

https://github.com/google/certificate-transparency-rfcs/pull/173

Regards,
Andrew

_______________________________________________
Trans mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/trans

Reply via email to