From: John Stanley (stanley@peak.org)
Date: Thu May 01 2003 - 12:27:12 CDT
Kai Henningsen (kaih@khms.westfalen.de):
> It is, however, an important part of interoperation in Usenet.
Hardly. There is a specific header designed for this purpose, and agents
that ignore that header are broken. The author of the agent decided this
alleged "interoperation" wasn't important, so it's his problem.
And I say "alleged" because USENET will work just fine without any "Re: "
appearing anywhere in the Subjects. The "interoperation" straw man has
blown down a long time ago.
Charles Lindsey (chl@clerew.man.ac.uk):
>Yes, but why do you want to remove a feature, which is widely used in both
>News and Email ...
A. None of the news or email clients I used do anything with it, so I
doubt the claim that it is "widely used in news and email." About all they
do is insert it because they've been programmed to do so; they don't give
a damn if it appears in anything I'm reading or not.
A1. Given the propensity for spammers to start out their contacts with
subjects like "Re: the information you asked for", I question that it is
being used correctly in a majority of email today.
B. It was a hack in the first place. An ugly kludge that has long ago been
superceded by a specific header intended for this purpose.
C. If it is so vitally critical for users to see what things are replies,
then the agent can easily mark any message that contains a References
header, just as they seem able to mark messages that contain an Importance
header, or mark messages that contain Mime-Version headers as MIME
messages. My absolutely ancient copy of elm can do both of the latter.
Why can't news agents?
>And you want to replace it by a feature (the References header) which is
>not usually displayed by reading agents,
Well, point proven. It is was so critical to display, agents would be
displaying it now. In any case, HE doesn't want to replace something with
the References header, it has already been replaced -- by this very group.
>Far better to document the _correct_ way to use it (which 95% of current
>agents already seem to get right).
Well, if you like jaywalking, I suppose you'd say that it is better to
document the correct way to jaywalk than to actually stop people from
jaywalking.
I.e., I think the point being made is that it is NOT the correct way to do
this, the correct way to do this is to use the structured header already
defined for this purpose and not to coerce an otherwise unstructured
header into double duty.
>We
>are not writing standards for the benefit of people who deliberately write
>awkward cases that are not going to happen in real life.
But we ARE writing standards for people who refuse to use the headers
already defined for specific purposes?