ietf-nntp Articles without message IDs

[Charles Lindsey]

In <Hk1zamPQQfi+Ewje at romana.davros.org> "Clive D. W. Feather" <clive at on-the-train.demon.co.uk> writes:

>In message <yly92v5haw.fsf_-_ at windlord.stanford.edu>, Russ Allbery 
><rra at stanford.edu> writes

>> 2. State that articles MUST have message IDs, but add a special exception
>>    to the text of POST saying that articles offered via POST need not
>>    have a message ID, but that in that case one will be added by the
>>    server.

>We don't even need to do that. We've got the concept that an article has 
>a message-id, and therefore the server will need to create one. That 
>doesn't have to be from a message ID header, and so this doesn't need to 
>be called out.

The situation to be avoided at all costs is that two copies of the same
article on Usenet should have different message-ids. So it needs to be
clear that, if the server adds one, then that same message-id must be
attached to that article whenever it leaves that server (so, for exmample,
if is is passed on to another server via IHAVE (or even POST), then that
other server MUST see that same id, and not create yet another one).

There may be several ways to specify that in the draft. Saying that a
Message-ID header is to be added is one way to do it, and might be the
easiest way to specify it (have we already said somewhere that the
messag-id comes from the incoming Message-ID header, if present, by way of
that oracle?). I don't really mind so long as the effect is achieved
somehow.

>>Maybe "the server MAY synthesise a message-id and add it to the article
>>headers following whatever article format specification is used for the
>>articles it is serving" instead, to clear things up for OVER and to make
>>behavior match existing practice.

>I'm not sure about that. Elsewhen we said that servers should never be 
>fiddling with article contents. At the least we should word it to allow 
>the ID to either be added to the article or not.

We certainly shouldn't be saying that with regard to the POST command.
There are lots of things that an injector MUST/SHOULD/MAY do to an
article, as specified in RFC 1036 / Usefor / wherever. So you just say
that a POST command does whatever is customary to do when injecting an
article, and wave your hand in the direction of the relevant Netnews
standards (indeed such a wording is already present). With a bit of luck,
you can solve the message-id problem at the same time.

>>> * Does OVER return the message-id or the contents of the Message-ID header?
>>The contents of the Message-ID header, I'd say.

>Okay.

Yes. OVER is explicitly in the business of returning real headers.

>Do we want a standardised metadata item (":message-id" perhaps) for the 
>ID, unadorned by any comments or other syntactic silliness? This would 
>be something that servers could make available as well, but wouldn't be 
>required to.

No.

-- 
Charles H. Lindsey ---------At Home, doing my own thing------------------------
Tel: +44 161 436 6131 Fax: +44 161 436 6133   Web: http://www.cs.man.ac.uk/~chl
Email: chl at clw.cs.man.ac.uk      Snail: 5 Clerewood Ave, CHEADLE, SK8 3JU, U.K.
PGP: 2C15F1A9      Fingerprint: 73 6D C2 51 93 A0 01 E7 65 E8 64 7E 14 A4 AB A5