From: Bruce Lilly (blilly@erols.com)
Date: Wed Apr 28 2004 - 23:49:46 CDT
Charles Lindsey wrote:
> In <408EA552.2010902@erols.com> Bruce Lilly <blilly@erols.com> writes:
>
>
>>Charles Lindsey wrote:
>>
>>>In <4087D7E0.5020108@erols.com> Bruce Lilly <blilly@erols.com> writes:
>
>
>>>>There is no need for such contortions, as
>>>> Followup-To: poster
>>>>and
>>>> Followup-To: foo@example.com
>>>>are unambiguous. The first of those is legal and the second is less likely
>>>>to cause problems than your example.
>
>
>
>>>Anyway, the point is moot because neither could be introduced onto
>>>Usenet at the present time because of the need for backwards
>>>compatibility.
>
>
>>By the same argument, Mail-Copies-To cannot be introduced.
>
>
> Not so. The Followup-To-header is already defined and widely implemented,
> so we cannot suddenly introduce new syntax (though in a few years time,
> when agents know how to parse - and ignore - extension parameters, that
> might become possible).
>
> But that does not apply to the Mail-Copies-To header. Firstly, that header
> is already in use quite widely, and all our draft does is to confirm and
> tidy up that existing use. Secondly, the sky does not fall in if some
> agents do not at first recognize and act upon it (because that is
> precisely the current situation, and the sky is still up there).
You have utterly missed the point. Currently (RFC 1036) an article
containing the fields:
Reply-To: foo@bar.edu
Followup-To: poster
would result in a mailed followup to the specified address. Under the
November draft text (the only text that I can readily examine for
consequences, as it is the last text that you have made available in
full to the WG), that same article with the same fields would result
in *NO* followup at all, since there is no Mail-Copies-To field, and
the default in the absence of such a field means that "The followup
agent SHOULD NOT email a copy of the followup to the poster".
That's silly; what's the point of "Followup-To: poster" is the
*absence* of a completely different field causes *NO* followup at
all to be sent. Even sillier is the legal set of fields:
Reply-To: foo@bar.edu
Followup-To: poster
Mail-Copies-To: nobody
Whether or not the field is currently used (w/o definition of either
syntax or semantics" and whether or not "the sky is still up there"
are beside the point. This is a Working Group of the Internet
Engineering Task Force, and it is our responsibility to make sure
that sound engineering principles are followed. If that means
revising a half-baked scheme, then so be it. As it stands, there
are conflicts between Mail-Copies-To and Followup-To. Moreover,
indicating that followups should be sent to newsgroups as well as
email addresses involves multiple header fields, which runs the risks
that
a) one of the fields might be lost, resulting in changed semantics
b) different followup agents are likely to respond differently.
The latter is probably inevitable to some extent with new functionality,
but the former is evidence of poor design and that in turn is a
"known technical omission".
I have proposed a solution that addresses the problems above. I do
not think that it is possible to "fix" the conflict between
Followup-To and Mail-Copies-To while maintaining two separate
fields -- I believe that the only way to address the issue is to
combine the functionality in a single field which indicates where
followups are to be sent (both newsgroups and email addresses). I
am sure that such a solution is possible, as I have already implemented
it. You are of course free to propose an alternative, but please
carefully consider the issues before doing so (preferably by
implementing your proposal off-line) so as not to waste WG time.
>>>What conflicts? I have already fixed the minor problem that you raised
>>>earlier (see the new 8.6 Step 5).
>
>
>>What "new 8.6 step 5"? The only 8.6 that I have access to is dated November
>>2003, so can hardly be called "new".
>
>
> I explained the small change I made to fix your problem at the time that
> you first raised it, and the recently circulated text regarding back
> references just happens to include that, since it contains the whole of
> 8.6 as it then stood. And sorry, it is not a new Step 5; is is just a
> one-word change to the former Step 5.
With no text available and no reference to where the text might be found
I can only add (to my remarks above to the effect that no "small change"
is going to address the issues) that I again support John Stanley's call
for you to make the full text of the draft as you think it stands available
to the WG so that we can have fruitful discussions -- we cannot do you if
you are the only person who has access to the text.