ietf-nntp HDR replacement text

[Russ Allbery]

Clive D W Feather <clive at demon.net> writes:
> Russ Allbery said:

>> 9.5.3  The HDR Extension

>>   If the header occurs in a given article multiple times, only the value
>>   of the first occurrence is returned by HDR.

> Is that too prescriptive ? Would it be better to say "only the value of
> one of the occurrences is returned" ?

My preference is to fully specify the protocol rather than leaving
anything undefined, but I could be convinced that's a bad idea for some
reason.  NNTP to date has preserved header order, however, and I think
it's good to encourage software authors to continue to do so as it may be
potentially useful for things like header signing.  And a fully specified
protocol is always nicer when possible.

>>   If the requested header is present but empty in a given article, a
>>   line for that article is included in the output but the header
>>   content portion of the line is empty.  If the requested header is not
>>   present in a given article, no line for that article is given in the
>>   output.

> Is that consistent with other commands ? I seem to recall that we
> decided that OVER would treat no such header and empty header as
> equivalent.

I'm not sure there's much else to be consistent with.  The difficulty with
OVER is that there's no real way of indicating an empty header; HDR has a
very good way, so why not use it?

>>   If the optional argument is an article number and no article with
>>   that number exists in the current group, a 423 error response shall
>>   be returned.

> I don't think this is the right thing to do. Consider:

>     HDR subject 12345
>     HDR subject 12345-12345

> Shouldn't these always return the same thing ? If there's no article
> 12345, what will/should the latter do ?

> Compare it to:

>     HDR subject 12345-12999

> when none of the articles in that range exist.

Hm.  Good point.  Having it just return an empty result is consistent with
how XHDR is currently implemented in INN and also seems consistent with
XOVER.  It's *not* consistent with the OVER text currently in the draft,
however, which states:

    If no articles are in the range specified, the server returns a 420
    error response.

I believe 420 is the wrong error code, though; this should also be 423.

I prefer that all of your examples above do the same thing, and that all
of them return a 423 error code.  Even more important than that, however,
is for HDR to do the same thing that OVER would in the same situation.  I
think producing an error code rather than an empty response is cleaner and
clearer for the client.

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