[NNTP] Article Numbers Becoming Invalid (RFC 3977)

[Julien ÉLIE]

Hi Russ,

>>   If the current article number is already lower or equal to the lowest
>>   existing article number (if any) in the newsgroup, a 422 response MUST
>>   be returned.
>
> Shouldn't we go ahead and use the same terminology?  Something like:
>
>    If the current article number is invalid, a 420 response MUST be
>    returned.  If the current article number is equal to or lower than the
>    reported low water mark, a 422 reponse MUST be returned.  [...]
>
> This also reverses the order so that it's clearer that an invalid article
> number takes precedence in determining the return code.

The change of order is indeed better.

With the wording using the low water mark, I have doubts whether people
won't get confused.  It implies that the low water mark is moving without
actually being *reported* (after GROUP, the server says 12 so it is the
reported low water mark; and if article 12 is cancelled, then the low
water mark is 13 but the server does not mention it to the client).

Maybe that's OK to use "reported low water mark" even in that situation,
but it is something that is not done elsewhere in RFC 3977.

Couldn't we mention that it is the reported low water mark at the instant
of the execution of the command?


But I believe there is a major issue then:  it breaks the as-if principle
because we will end up with something unspecified for news servers that
cannot cope up in real time with the low water mark exactly matching
the first existing article in a newsgroup...

reported low water mark = 12
first existing article number = 13

GROUP -> current article number = 12
ARTICLE -> absent (423 response code)
NEXT -> current article number = 13
LAST -> ??

I think it should really be the "lowest existing article number"...

-- 
Julien ÉLIE

« Il vaut mieux un tapis persan volé qu'un tapis volant percé ! » (Astérix)