From: John Moreno (phenix@interpath.com)
Date: Wed Sep 22 1999 - 19:27:03 CDT
chl@clw.cs.man.ac.uk (Charles Lindsey) wrote:
> phenix@interpath.com (John Moreno) writes:
>
-snip-
> >> could we seriously consider adding the Mime-like "usage=..." to
> >> the present Supersedes header? Yes that is not a nice thing to do
> >> to a well-established existing header, but has anyone checked
> >> whether any existing implementations would actually barf on it?
>
> >I've just checked two servers, INN 1.7.2 and RumorMill 1.2, both
> >barfed (more precisely they didn't work, but the articles were
> >accepted which is even worse).
>
> This seems contrary to what Russ was saying, but I think it indicates we
> cannot risk tinkering with the existing Supersedes.
No, it's consistent with what Russ says -- the message itself goes out
normally, but the Superseding never matches because it looks at the
full header and/or requires the full header to be syntactically correct
for a message id.
So if you put anything besides the message id in the header Supersedes
breaks and both articles get propagated.
> >> >> 3. Simplification of Content-Type discussion for Mime,
> >> >> addition of Content-Disposition, and replacement of
> >> >> application/news-transmission by message/news (to pacify the
> >> >> Mail gurus in IETF).
> >>
> >>
> >> No, I think the proper thing to do is to declare what the default
> >> is if it is absent. You will see that I was rather vague on what
> >> the default should be, since current practice is divided. I was
> >> hoping that one of you would stick your neck out and recommend a
> >> more definite default for us to specify.
>
> >I'd say attachment if not included, but I'm not big into MIME.
>
> The you will find that the Turnpike people take the opposite view, I
> think :-( .
>
I think that someone will take the opposite view no matter what we do,
I still feel that it's the right thing to do (as well as feeling that a
proper design would include a toggle for it).
-- John Moreno