ietf-nntp Article concepts

[Russ Allbery]

Clive D W Feather <clive at on-the-train.demon.co.uk> writes:

> What I'm trying to say is that a client mustn't barf if it finds an
> article that conforms to this but not 1036 or 2822.

> For example, if the reply to ARTICLE contains a CR not followed by LF,
> the client can reasonably close the connection immediately because the
> server has broken a MUST NOT. On the other hand, if the reply begins:

>      Headers: This does not conform to 1036 or 2822.

>      This is the first line of the body ...

> it has to cope with it. It might decide not to present the article to
> the user, but it MUST NOT go wrong.

Hm.  I'm not sure what this practically means for a client implementor.
The client can always just immediately close the connection; that's not
really something we can restrict.  It feels like this is a quality of
implementation issue more than really a protocol issue.  But I don't have
a good suggestion for alternate text, and don't really mind leaving it
alone.

> Picked nit: I don't say that the body must end with CRLF, I say that the
> article must end with CRLF. If the body is empty, this CRLF is the one
> at the end of the separating line.

Ah, yeah, good point.

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