active file too far ahead

Per Hedeland per at erix.ericsson.se
Wed Oct 27 17:48:09 UTC 1999


Kjetil Torgrim Homme wrote:
>
>[Brian Kantor]
>
>>   GROUP should return the highest article number which has ever
>>   existed, even if that article has subsequently been removed due to
>>   cancel, expire, or supercedure.

If you say so:-) - but *why*? The reasons not to lower the himark that
is the basis for future article number assignments (i.e. the one in the
active file) are obvious of course, but there's no obvious (to me:-)
reason that this dictates anything at all about the GROUP response.

>If so, INN 2.2 is broken.

As broken as most NNTP implementations since day one, then. Both the
"reference implementation" and INN 1.x simply report the result of a
directory scan on the GROUP command. Or rather, they go to the extra
trouble of doing a directory scan, just to be as "truthful" as possible
in the GROUP response.

>  Per Hedeland doesn't agree with you, he
>says my Gnus is broken since it relies on output from GROUP instead of
>LIST ACTIVE.

Well, perhaps not exactly "don't agree" (might be a stupid thing to do
when talking to the RFC author about how the protocol works:-) - it may
well be that the GROUP response "should" report the highest himark that
has ever existed (i.e. basically the one from the active file), even
though no implementations do so and there is no particular reason for
it.:-)

But your Gnus should definitely be able to deal with GROUP's himark
being lower than highest-ever/active's - and I'm sure it does, since
otherwise it wouldn't work with the vast majority of NNTP servers out
there. What problems are you actually having with the GROUP behaviour
you see?

>>   I hoped I'd made that clear the first time :-), but I'm fairly
>>   certain it has also been clarified in the NNTP Common Practices
>>   RFC.
>
>(which one is that?)

Me too...

>draft-ietf-nntpext-base-08.txt says:
>
>        No similar assumption [refers to lowmark which can never
>        decrease] can be made about the high water mark, as this can
>        decrease if an article is removed, and then increase again if
>        it is reinstated or if new articles arrive.

This definitely agrees with actual reality, at least.

>I think this mandates INN's behaviour...

s/mandates/allows/ and I agree.

--Per


More information about the inn-workers mailing list