Re: [dispatch] please dispatch draft-bhjl-x509-srv-02.xml

["John Levine" <[e-mail]>]

More review is nice, but useful review needs to be more than "I don't
trust this."

>Primarily, there needs to be adequate security review.  There are
>several big risks that this draft skirts and good review, and maybe
>even analysis, of this is important.

This draft describes a way to publish and retrieve S/MIME and PGP
keys.  The keys are the exact same ones you can get in other ways now,
and the draft goes to considerable lengths to say that, just like with
keys you get any other way, it's up to the client to decide how much
credence to give to them.  Any security problems with bogus keys are
the same ones we've already been dealing with as long as there have
been PGP and S/MIME.

If it appears that this draft is asserting that the keys are "real"
because the domain says so, it's trying very hard not to say that, so
it'd help to know what was confusing.  (Attempting to specify the
relationship between a domain and its users is a hopeless swamp of
despair where we're not going.)

>The draft appears to violate RFC 7320.  That suggests that wider
>review of other areas of this is needed.

I presume this is a reference to the URL paths in section 3.  They are
imported verbatim from RFC 4387 which was published ten years ago.
Since they're not new, it would be helpful to clarify under what
circumstances we're not allowed to use specs from existing standards
track RFCs.

R's,
John