ietf-822
[Top] [All Lists]

Re: Draft for signed headers

1999-03-25 05:06:56
In <19990324165941(_dot_)12273(_at_)main(_dot_)templetons(_dot_)com> Brad 
Templeton <brad(_at_)templetons(_dot_)com> writes:

The reason they perhaps ignore it is they already have a couple of standards
on the table, with working implementations in some cases.  These include
S/MIME and multipart/signed, as well as a PGP form.   The chances of them
taking another as a draft are slim to nil.

Sure, but those "couple of standards" are also useable in news (agreed the
S/MIME one is grossly unsuitable, but the PGP one is fine for news). So
themail position is not really different from ours. They could usefully
use an additional headers signing mechanism just like we could, except that
our need is far more pressing.

But since there are a lot of big advantages to a signed header scheme (along
with some disadvantages when it comes to mail) if USENET adopts one it is
entirely possible that some mail tools would also adopt one, especially
any merged tools.

Agreed. So it would be sensible that any scheme we devise should be
workable in mail as well. That limits us a bit, but not too much, as my
Draft was intended to show.

USENET and mail have some needs in common, but a few big differences
will make unification unlikley at present:
      
      a) Mail doesn't care nearly as much if the signatures are bulky
         and the certificates are bulkier.   Typical signed E-mail
         certs in S/MIME with X.509 can be multiple kilobytes -- the size
         of the average USENET post.

Sure, but how much is S/MIME actually used in email with those
certificates? For example, Lotus Notes is widely used, but doesn't use
multiple kilobytes (yes, I know it has a LEAK in it for the benefit of the
NSA, but that is by the way). But clearly, any header signature mechanism
would not be based on that. The headers added by pgpverify are just fine.

      b) Mail has pretty fully moved to MIME support.

And we are pushing News in the same direction. But please note:

I AM NOT TRYING TO PROPOSE HEADER SIGNING AS THE NORMAL AUTHENTICATION
METHOD FOR USENET. That is your interpretation, not supported by others. I
am proposing it for specific applications (newgroups, cancels, etc, with
more to follow when demand arises). I am quite content for it to coexist
with existing MIME-based methods.

      c) With mail  you know of and control who receives your mail.  Thus
         mail can happily support multiple competing methods, as long as
         the sender and recipients of each *specific* mail message
         understand the signature and certificate space.   With USENET,
         we have the "Highlander" problem -- "There can be only one."

No. The requirements of mail and news are similar here. You do not want
too many methods in either medium - you want standardised methods
supported by many agents in both cases to promote the maximum
interoperability. But OTOH "not too many" does NOT equate with "only one".

      d) With mail, verification will be done by MTA or client (however
         MTA is safest) but there are no intermiediate points that want
         to verify, unlike news where all relayers SHOULD verify.  So there
         is less load problem and more choice.

I disagree totally.

-- 
Charles H. Lindsey ---------At Home, doing my own thing------------------------
Email:     chl(_at_)clw(_dot_)cs(_dot_)man(_dot_)ac(_dot_)uk  Web:   
http://www.cs.man.ac.uk/~chl
Voice/Fax: +44 161 437 4506      Snail: 5 Clerewood Ave, CHEADLE, SK8 3JU, U.K.
PGP: 2C15F1A9     Fingerprint: 73 6D C2 51 93 A0 01 E7  65 E8 64 7E 14 A4 AB A5