Re: [Question] What is an accurate meaning of "A server MAY cache a dead server indication"?

[Paul Vixie <paul@vix.com>]

# > that's probably why it says "OPTIONAL".
# 
# Yes, but the section also uses "MUST".  If a "MUST" is optional, it is
# better written as "SHOULD".

agreed, i think.  what the combination means (to me) (in this case) is that
if you don't want to have dead/unreachable server logic, you don't have to,
but if you do want it, it has to work like this.  that makes no sense, and
should have been written with a combination that added up to "if you want 
this particular method of dead/unreachable, here's how to do it".  padlipsky
tried hard to shine light on the "descriptive vs. prescriptive" approach to
protocol design, but apparently not everybody read his book.

fortunately, rfc's can be read very loosely, and the ultimate arbiter of an
implementor's success is interoperability, not conformity.  so if various
folks through the years come up with alternative dead/unreachable server
logics that they either do or don't publish as rfc's, then each one will
succeed or not based on whether it improves the operation of a resolver, and
(perhaps) on whether it degrades the performance of other resolvers/servers,
but never whether it exactly matches what's written in section 7.2 here.

i'm glad our WG chairs are now actively searching for ways to make sure
nothing gets to IESG without strong support and actual review.  this doc,
among many others we've sent up, had neither.  (mea culpa, i helped w/ this.)

--
to unsubscribe send a message to namedroppers-request@ops.ietf.org with
the word 'unsubscribe' in a single line as the message text body.
archive: <http://ops.ietf.org/lists/namedroppers/>