On 23 Sep 2000, Greg Stark wrote:
> Stephen Turner <[EMAIL PROTECTED]> writes:
>
> > As to the question above -- I think Reply-To: list is actually more common.
>
> Setting Reply-To on a mailing list is simply incorrect behaviour.
Actually, there's one big reason why I disagree with this: the document
completely omits to mention /the most common/ list reply mechanism, and
makes out that its suggestion is a panacea, which it isn't.
The most useful reply mode is `to list and no one else'. Second useful
after this is `to list and sender'. Third is `to sender only'.
In the non-munged model, there's no easy way for the mailer to distinguish,
from the standard 822 headers, the list reply address. Therefore you
only get a choice of the second- and third-most useful mechanisms;
if you want to reply to the list alone, you have to manually go through
and trim all the headers -- tedious over slow links.
This means that everyone ends up being lazy and simply sending their
reply to /everyone/ on the From: and Cc: line, meaning, typically, that
dozens of people end up with junk threads in their mailbox no longer
relevant to them. This is the case on almost every single list I use.
It also has other horrid knock-on effects: discussions which really
should be on the lists end up happening as a private exchange, because
at some point down the line some of the contributors hit `reply to sender
only', after which point they're almost inevitably locked in to a small,
two- (or perhaps more) person conversation: this means the list members
often miss the important machinations of a rationale being forged.
This is also the case on almost every single list I use (the Debian lists
being numerous).
How often do people really want to reply /only/ to the sender? Rarely.
If mail clients ubiquitously had a `reply to sender' option that was as
easy to select as `reply to from', `reply to reply-to' and `reply to
all', the problem would be solved. However, no mailer I've come across
has this.
<phew>
> (i) vacation programs should use the envelope sender, they shouldn't look at
> _any_ header at all for the return address.
>
> (ii) they should ignore messages with Precedence bulk, where the user isn't
> listed in the recipient headers anywhere and whatever other methods for
> detecting mailing lists can be devised.
>
> (iii) Such programs should keep a cache of already responded to addresses to
> avoid responding to the same person twice.
>
> (iv) Such programs should be able to recognize their own responses and not
> respond to itself.
Yes. Corporate software is generally astoundingly, mind-numbingly,
horribly backward. In the Free software world, programs /tend/ to get
dropped if they're that bad. :-)
> Personally I recommend banning any messages posted from Novell Groupwise 5.2
> from posting to the list now. The vacation program he used is presumably built
> in to that client and it's only a matter of time until someone else tries to
> use it.
While it's tempting to be hasty and draconian, remember that the mailer
does not represent the person, and the person does not represent the
mailer, and, in mitigating circumstances, the link between them may be
immutable.
c.
------------------------------------------------------------------------
This is the analog-help mailing list. To unsubscribe from this
mailing list, send mail to [EMAIL PROTECTED]
with "unsubscribe" in the main BODY OF THE MESSAGE.
List archived at http://www.mail-archive.com/analog-help@lists.isite.net/
------------------------------------------------------------------------
