On Thu December 2 2004 18:58, Steve Dorner wrote:
Re: Attempts at establishing harmful conventions
Date: 2004-12-02 18:58
From: Steve Dorner <sdorner(_at_)qualcomm(_dot_)com>
To: Kai Henningsen <kaih(_at_)khms(_dot_)westfalen(_dot_)de>
CC: ietf-822(_at_)imc(_dot_)org
At 8:27 PM +0200 12/2/04, Kai Henningsen wrote:
> For that matter, how many permit a user to select multiple
messages to compose a response citing all of the selected
messages (including appropriate setting of In-Reply-To and
References fields)?
Hopefully few, as there *is no* appropriate setting for those fields in
that situation.
We actually hashed out a way that seemed right enough on this list a
year or so ago.
There was a discussion in June 2003, however if it can be
said to have reached any conclusions, I believe those
conclusions include:
1. imposing new syntax on the References field is not
backwards compatible
2. expecting UAs to coordinate updates to multiple fields
is probably over-optimistic
3. several existing UAs botch References and In-Reply-To
fields, destroying their utility
4. repeating message-ids may make a field "too long" (there
is of course no limit to the length of a logical (continued)
field, however a msg-id doesn't look like "1", it looks
more like
"<supercalifragilisticexpialidoceous(_at_)xn--blurfl(_dot_)xn--grimble(_dot_)xn--pritz(_dot_)foo(_dot_)bar(_dot_)com>",
(that's a short example -- the domain name on the RHS can be
up to 255 octets, and the local-part can also be lengthy)
and one wouldn't want to have to repeat that)
5. rather than overload additional semantics on the References
field, a new field or set of fields should be developed
So far, I haven't seen any detailed proposal to address the
issue in accordance with those conclusions (indeed, several
points are somewhat contradictory).