From: John Moreno (phenix@interpath.com)
Date: Thu Jul 16 1998 - 10:49:12 CDT
Greg Berigan <gberigan@cse.unl.edu> wrote:
> phenix@interpath.com (John Moreno) wrote:
> >Greg Berigan <gberigan@cse.unl.edu> wrote:
> >>phenix@interpath.com (John Moreno) wrote:
> >>>Greg Berigan <gberigan@cse.unl.edu> wrote:
>
> >>>> I am not suggesting that the "Author-IDs" [1] header will be immediately
>
> >>> I agree - as long as it's not being proposed as solution for a existing
> >>> problem. [...] (of course the name should be changed to P-Author-IDs so
> >>> agents which don't know about it, but do follow the draft can implement
> >>> it).
>
> >> Agents which follow the draft will know about it and won't need the P-
> >> prefix to act as a crutch to tell them how to do it.
>
> > I would suggest that it should be a X-P header for the moment (also -
> > it's not likely that agents will immediately start implementing all of
> > the features, this get's it in as soon as the header classes are added).
>
> Are you saying it is acceptable for agents to ignore parts of the draft
> while implementing others?
No, I'm saying this *will* happen. Different thing entirely. Also some
sections *will* be overlooked - and if they aren't providing a way of
setting it then this will be one of them.
> Are you then not also saying that User-Agent should be in the draft as
> X-User-Agent? That Mail-Copies-To be in the draft as X-Mail-Copies-To? I
> thought it was the design that no X- headers be directly in the draft. I'm
> only extending that to any prefix, as once it is in the draft, agents are
> expected to support it.
I think what I was groping for is that we should include a reference to
existing prefixed headers at the time the draft is released, and that
this be one of those headers. As for Mail-Copies-To being in that
category, possibly - it's already being used is the main objection I'd
have to that. As for User-Agent - it absolutely should NOT be in that
category, it has already been tested as X-Mailer, X-Newsreader and so
forth and we are now moving it into permanent use as User-Agent
(consistent with the previous policy of X never being used for headers
which are intended to be adopted).
-snip-
> Newsreaders that support the P- and X-P- prefixes should also support the
> Author-IDs header as persistent out of the box. Newsreader authors which
> adopt the new standard piecemeal should be shot.
Whether they should or not is irrelevant - they will. They will also
implement most of it and miss part of it, not through any deliberate
decision but through sheer incompetence (in connection with the GNKSA
I've been asked by "authors" 'what's References and where can I find it
in the RFCs', and although it's optional I don't think In-Reply-To has
been left out of so many mail agents because they didn't like it, but
rather because they missed it).
-- John Moreno