ietf-822
[Top] [All Lists]

Re: Understanding response protocols

2004-09-23 10:53:41

Claus Färber wrote:
Hallo,

Bruce Lilly <blilly(_at_)erols(_dot_)com> schrieb/wrote:

Claus Färber wrote:

Further, it does not allow something like this:
| From: jdoe(_at_)example(_dot_)com
| To: product-announce(_at_)list(_dot_)example(_dot_)com, 
product-discuss(_at_)list(_dot_)example(_dot_)com
| Reply-To: info(_at_)example(_dot_)coom
| Mail-Followup-To: product-discuss(_at_)list(_dot_)example(_dot_)com


"No, it does not work with existing UAs." Most UAs ignore illegal
fields such as "Mail-Followup-To".


When discussing a new potential standard, it does not make sense to
slanderously refer to proposed fields as "illegal". They field is
currently undefined but the idea of introducing a new spec is to change  
that.

1. I'm not aware that we're discussing any "new potential standard";
   as Keith has pointed out, this list is not associated with any
   active WG, therefore has no charter, etc., and the present
   discussion is not related to any proposed draft.
2. Fields which are not defined in any RFC are referred to by at
   least one UA's author(s) as "illegal", so use of that term is
   appropriate in the context of UAs ignoring non-standard
   fields -- that's how they're viewed by (some) UA authors.
3. The point remains that "Mail-Followup-To" is ignored by the
   vast majority of all UAs and by all of the most widely used
   UAs.  I.e. you initiated the discussion of what does and
   does not work with existing UAs -- and "Mail-Followup-To"
   does not work with any substantial fraction of existing UAs.

And the following achieves what I believe you intended using only
standard header fields, with (as discussed) the From field for
narrow responses, Reply-To for default responses, and a user-edited
selection from all addresses for a wide response:


From: info(_at_)example(_dot_)com
Sender: jdoe(_at_)example(_dot_)com
To: product-announce(_at_)list(_dot_)example(_dot_)com, 
product-discuss(_at_)list(_dot_)example(_dot_)com
Reply-To: product-discuss(_at_)list(_dot_)example(_dot_)com


A user agent following established practice

What happened to "when discussing a new potential standard" -- are
you trying to have your cake and eat it too?

will send narrow replies to
"product-discuss(_at_)list(_dot_)example(_dot_)com" (Reply-To) and broad 
replies to
"produxt-announce(_at_)list(_dot_)example(_dot_)com, 
product-discuss(_at_)list(_dot_)example(_dot_)com"  
(To, Reply-To), which is also wrong.

(Several) existing UAs provide a "reply-to-author" feature which
uses the From field, ignoring Reply-To, which happens to be
what Keith has suggested should be done.  Such narrow responses
go to "info(_at_)example(_dot_)com". Is that not where you expect narrow
responses to go?

Broad responses as discussed here almost invariably require
hand editing, so go to an edited set of mailboxes.

IOW: Both solutions are not backwards-compatible wrt broad replies but
your proposal (it's not the existing standard because it adds additional
semantics to the existing header fields)

No -- there are no additional semantics. The From field still
holds the mailboxes of the author(s), and reply-to has exactly
the semantics specified in RFC 2822, viz.:

   When the "Reply-To:" field is present, it
   indicates the mailbox(es) to which the author of the message suggests
   that replies be sent.

also breaks narrow replies

No, narrow responses as discussed on this list go to the author(s)
[as specified in the From field as defined in RFC 2822 and its
predecessors].

(and
Mail<->Netnews gateways).

No, since the fields in question are all standard fields and have exactly
the same unchanged semantics in both mail and news messages.

Please note that you can't put two addresses in the From field either,  
in case that the message has been jointly written by two authors (e.g.  
"jdoe(_at_)example(_dot_)com" and "jane(_at_)example(_dot_)org").

Incorrect, as pointed out in detail elsewhere.