Always 224 with XOVER?

[Russ Allbery]

Julien ÉLIE <julien at trigofacile.com> writes:

> Instead of writing such a page (which can be very large and boring with
> texts available elsewhere), wouldn't it better to write a page with
> *only* examples?  I reckon that being practical is far better than RFC,
> especially for telling how to behave in (all?) cases, and also
> differences.  Very few comments.

The idea that I had in mind was a sort of annotated guide to both practice
and the standards, sort of a commentary on the standards text, but I only
had a vague idea in mind and hadn't gotten anywhere concrete with it.

> The page will also be very large :)
> But maybe it could be split:  one page by NNTP command?

Oh, yes, that was my plan, and maybe a page per response code, and a few
other different ways of organizing hte information.

> For each command, we would have a link to the general AUTHINFO page and
> explanations if something is different here (not 480/483/502).  Then
> several cases which show answer codes and subtleties.
>
> Example for XOVER:
>
> * Most news servers return 224 even though no articles matches:
>
> GROUP existing.newsgroup
> 211 11 10 20 existing.newsgroup
> XOVER 1-8
> 224 No articles in 1-8
> .
>
> Even though it contradicts RFC 2980 which defines XOVER (423 should be
> returned in such cases), implementations usually return 224 so it is
> best to keep returning 224.
> Note that OVER must return 423 in such cases.

Yes, this sort of thing would be great.  Basically, all of the answers to
the discussions we've had recently about NNTP standards.  Your examples
are all excellent.

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