[NNTP] Article Numbers Becoming Invalid (RFC 3977)

Russ Allbery rra at stanford.edu
Mon Jan 4 10:17:22 PST 2010


"Charles Lindsey" <chl at clerew.man.ac.uk> writes:

> But it seems we are still expecting GROUP to report the best LWM available
> to that particular server at that instant, which may be higher than what
> LIST ACTIVE just reported (and may be out of date again 10ms later :-( ).

RFC 3977 defines all of the occurences of low water mark by reference to
GROUP, so I think it's dubious to have different commands use different
definitions.

>> Yes, or the result of GROUP would be inconsistent with LIST ACTIVE, and
>> specifically LIST ACTIVE could report a lower number than a previous
>> GROUP command, which isn't allowed.  (It's not allowed for reasons that
>> don't really apply here, but still.)

> But it is clear just that is going to happen with all reasonable
> implementations, so we need to ask whether it really matters.

Actually, right now INN reports the same numbers with LIST ACTIVE and
GROUP, so this doesn't happen with all reasonable implementations.  You
can avoid it by using the active file as the reference for all low water
marks.

> What evil then ensues? AFAICS nothing, because a subsequent GROUP will
> always report something not less than the previous GROUP. So what are
> the "reasons that don't really apply here", and can we think of a
> plausible excuse for ignoring them?

The news client is allowed to assume that no articles below the reported
low water mark will ever reappear, which is semantically equivalent to
saying that the low water mark will never decrease if the low water mark
is actually meaningful.  The harm (clients thinking articles no longer
exist on the server that actually do) doesn't apply in this situation
since the inconsistency would be due to the LIST ACTIVE command lagging,
and those articles wouldn't actually be accessible.

-- 
Russ Allbery (rra at stanford.edu)             <http://www.eyrie.org/~eagle/>


More information about the ietf-nntp mailing list