ietf-nntp Article concepts

[Matthias Andree]

[Resend from proper address]

"Clive D.W. Feather" <clive at demon.net> writes:

>    Each article SHOULD have a message-id, which is the value of the
>    header with name "Message-ID". Two articles processed by an NNTP
>    server MUST NOT have the same message-id. Note that RFC 1036 [6]
>    further requires that message-ids are globally unique for all time.
>    If an article contains more than one such header line, the server
>    MUST choose one as the message-id it uses (for example, it MUST NOT
>    allow the article to be retrieved using both message-ids). If an
>    article does not contain this header line, the server MUST NOT
>    allocate a message-id, and MUST use the placeholder "<0>" (without
>    the double quotes) where it is required to provide the message-id of
>    that article.
>
>    OUTSTANDING ISSUE
>
>       I would much prefer us to simply say there MUST be exactly one ID,
>       but that negates the existing text which says "<0>" MUST be used
>       if there is no message-id header line.

I'd second requiring exactly one message-id, as this is backed by
RFC-1036 and the current USEFOR draft anyways. NNTP isn't getting
anywhere close to resolving multiple-message-id conflicts and allowing
for articles without message-id and introducing the <0> special case
adds asymmetries, because articles without message-id can then only
be retrieved by their number in the respective group, but not by
message-id, and they don't show in LISTGROUP replies.

 Granted, many clients usually request by article number, but this
"<0>"-without-Message-ID thing might confuse existing clients or users.

-- 
Matthias Andree