Skip site navigation (1)Skip section navigation (2) 
Date:      Sat, 23 Nov 2002 02:38:44 -0800
From:      Terry Lambert <tlambert2@mindspring.com>
To:        Anthony Atkielski <anthony@freebie.atkielski.com>
Cc:        FreeBSD Advocacy <freebsd-advocacy@freebsd.org>
Subject:   Re: FreeBSD on the desktop (was: TheRegister article on Hotmail)
Message-ID:  <3DDF5AB4.E71E4E67@mindspring.com>
References:  <20021121161453.GA69019_submonkey.net@ns.sol.net>   <20021122234047.GB60785@wantadilla.lemis.com> <014201c29296$f9cc4a20$0a00000a@atkielski.com> <20021123023624.GA97416@gothmog.gr> <017101c2929b$18ef7e50$0a00000a@atkielski.com> <20021123033041.GA3884@gothmog.gr> <019901c292a3$b9c31690$0a00000a@atkielski.com> <20021123040925.GB4320@gothmog.gr> <01a201c292a7$23338d50$0a00000a@atkielski.com> <3DDF2A12.31DABCF5@mindspring.com> <01dc01c292c6$e31cca90$0a00000a@atkielski.com> ¤æ¿¿ìë¿¿ü¿¿á      eI	(øf	(   "P	(    æ¿¿CSß;pæ¿¿Àå¿¿ ±(                                                           Èð   `ç¿¿   ³ð   0À	                                   øå¿¿            æ¿¿Ìæ¿¿?(tæ¿¿¿ð8ç¿¿u(?® (àÑ (Dç¿¿á (Ò (   à  aÒ(ç¿¿           àÑ (=   üæ¿¿6Ò(?® (àÑ (Ò ( (ìë¿¿ÜQç¿¿¹¤(ç¿¿           àÑ (=   ¿ð?¤(ìë¿¿lQìë¿¿        (ç¿¿   t (   	            §P lü¿¿ÿ¿ðì¿¿ <3DDF4582.91B6820B@mindspr! ing.com> <021601c292d8$97fc49c0$0a00000a@atkielski.com>
next in thread | previous in thread | raw e-mail | index | archive | help 
Anthony Atkielski wrote:
> > Why?
> 
> Because the sender is one of the "all" to which the "Reply all" is sent.

Duplicate supression is a function of the receiver.

> > So that if the sender is not a subscriber to
> > the list, they don't se your response to their
> > message?
> 
> If they were not subscribers to the list, they would not have seen the
> message to which they are replying, and to which I am responding in turn.

Not true.  A frequent use of mailing lists is to:

1)	Have a problem.
2)	Search list archives for a solution to the problem.
3)	If a solution is found in the archives, STOP.
4)	If there is no solution in the archives, but there
	are other people with the probem, contact them directly,
	and see how they solved it (if they did).
5)	If the other people found a solution, and they share it
	with you, STOP (Note: subsequent people will repeat this
	process, until the people with the solution post it to
	the list to avoid being pestered).
6)	Post a request for a fix to a mailing list *without
	necessarily subscribing to the list*
7)	Receive an answer to the question, which is sent both to
	them directly, and o the list, for the purposes of being
	archived for future people with the problem.

Note that in step 7, the respondant sends to the list and to the
original sender, both, without the orignl sender being on the list.

In the case of cross-posting, etc., it's possible to have a set
of lists [A,B,C], and a set of users ith list membership a[A,B],
b[A,C], c[B,C], all involved in the discussion, due to an
initial cross-post.


> > It is the job of the receiving system to
> > perform duplicate suppression.
> 
> I guess my FreeBSD UNIX receiving system doesn't do its job, then.

I guess you haven't configured it correctly, then.  If you are using
POP3 against a FreeBSD mail server, then configure it to use procmail,
and maintain a rolling archive of "Message-Id:" field contents, and
suppress duplicate deliveries, on a per-user basis:

	# weed out duplicates
	:0 Whc: msgid.lock
	| formail -D 8192 $PMDIR/msgid.cache
	:0 a:
	$DELETE

See also:

	Procmail Mini-Tutorial by Tony Nugent
	http://www.sektorn.mooo.com/era/procmail/tony.html

It's generally accepted that this is actually the job of the MUA
(in this case, OutLook Express is failing to provide this feature
for you, for whatever reason; most likely, it's an overabundance
of the property "suck").  The reason for this is that suppression
of duplicates prevents you from greating MUA filter rules that do
proper archiving of messages and/or delivery into list archive
and standard mailboxes, as a result of their method of arriving
at the recipient (e.g. keying of the existance of a "Sender:"
or a "List-ID:" or "List-Unsubscribe:" or other header).


> > If it's not (it usually is not, for a myriad of
> > very good reasons), then it's the job of your
> > mail client and your mail client's filters to
> > deal with it.
> 
> If I am separately addressed and addressed as part of a mailing list, I
> _expect_ to receive two copies of the message.  The problem is with the
> sender specifying both me and the mailing list as recipients; I imagine many
> senders are using "Reply all" or its equivalent.

Your problem.  Not the sender's problem.

You could argue that it's also a mailing list manager issue,
since the mailing list manager could note that the "To:" or
"Cc:" line contained an address which mtched that of a known
list member for the message, and suppress delivery.  However,
this has two problems:

1)	The same problem with non-archival delivery that should
	be handled on behalf of the user by the MUA; at a bare
	minimum, this would have to be a subscriber controllable
	option, since most sae people *do not want the behaviour
	you want*.

2)	Doing this is computationally expensive, and the general
	rule in client/server software design is to push all the
	processing you possible can off onto the client, since
	clients outnumber servers by a wide margin, and the
	left-over 200MHz Pentium 2 box you are given to build a
	mail server with has far less balls than the 3GHz boxes
	sitting on the desks of the VP of sales, the VP of
	marketing, or the CEO's, CTO's, or even the technical
	writer's desk -- or even the 600MHz boxes they stick you
	with in engineering, because you don't "rate" a faster
	box, and since when is compiling the company's product
	as important as minimizing the browser containing porn,
	when someone walks into the sales manager's office?

If you are going to this much trouble to hack the crap out of your
list manager software, you'd be *much* better off addressing things
like the ability to send email to a list *minus* specific people,
and oher features that are requested a lot more than duplicate
suppression at the server instead of at the client.


> > Why do you have to do that?  THe "To:" or "Cc:"
> > field contains the list name.
> 
> No, it contains the address of the sender of the post to which I'm replying.
> The headers do not specify a "Reply-To:" address, and the "From:" field is
> filled with the address of the person who posted the message.  Thus, when I
> reply, the To: field is filled with the address of the poster, not the
> address of the list.  The error is not in the mail client, it is in the
> mailing list software.

The error is in the use of temil client, who used "Reply to
Sender" button, rather than t non-existant (because it's useless)
"Reply to List" button, or the "Reply to All" button that they
should have used, instead. 


> > I guess you need to spring for the full version
> > of Outlook to do that: cost of using a feature-poor
> > mail client.
> 
> I'm not aware of any "full version" of Outlook.

"Outlook", as oppoed to "Outlook Express".


> In any case, I should not wish to change the priority of headers in
> determining how the reply address is obtained.  The current behavior
> conforms to the RFCs, and that's fine with me.

Then lump it, because so does the mailing list management
software, and that's fine with everybody else.  8-).


> > You should probably point that out to the person
> > whose private email you sent to the list, in case
> > they are not a list subscriber and didn't see this
> > message, and appologize.
> 
> This is a discussion forum, not an exchange of love letters.

In other words, you are an unrepentant poster-of-private-email,
rather than an accidental poster-of-prviate-email.

Therefore the rest of us should consider that you will not
honor a confidence, before sending you private email.

-- Terry

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-advocacy" in the body of the message

Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3DDF5AB4.E71E4E67>