ietf-nntp AUTHINFO SASL protocol choices

[Andrew Gierth]

>>>>> "Russ" == Russ Allbery <rra at stanford.edu> writes:

 > Charles Lindsey <chl at clw.cs.man.ac.uk> writes:
 >> But if their upper bound is to be over that reasonable amount,
 >> then we shall have to define (and people will have to implement) a
 >> mechanism with an 'escape' to multiline, even though it is most
 >> unlikely it would ever be needed in real life.

 Russ> If the upper bound is over a reasonable amount, we should just
 Russ> always use multiline.  Let's please not provide two ways of
 Russ> doing something when we don't need to.

The reason for allowing an initial parameter is to avoid blowing a
round-trip unnecessarily.

e.g. if all the strings are <400-odd characters, for some hypothetical
SASL scheme in which the client goes first:

 C> AUTHINFO SASL method START client-string-1
 S> 3xx server-string-1
 C> AUTHINFO SASL method NEXT client-string-2
 S> 3xx server-string-2
 C> AUTHINFO SASL method NEXT client-string-3
 S> 2xx access allowed

(3 roundtrips)

whereas the same in multiline-only would have to be something more
like:

 C> AUTHINFO SASL method START
 S> 3xx continue authentication sequence
 C> client-string-1
 C> .
 S> 3xx server response follows
 S> server-string-1
 S> .
 C> AUTHINFO SASL method DATA
 S> 3xx continue authentication sequence
 C> client-string-2
 C> .
 S> 3xx server response follows
 S> server-string-2
 S> .
 C> AUTHINFO SASL method DATA
 S> 3xx continue authentication sequence
 C> client-string-3
 C> .
 S> 2xx access allowed

(6 roundtrips)

Having auth take twice as long would be a nuisance...

-- 
Andrew.