At 13.09 +0100 99-02-26, Charles Lindsey wrote:
Jacob Palme wrote:
If you have a hierarchical graph without loops, then all
new submissions should send to the top node, so lists below
the top node should only accept submission whose SMTP
sender is the top node!
No, I don't think that is right. Let us take an example. Both you and I
are on both of the lists ietf-822 and usenet-format, and we post to
each of those lists independently of the other.
Now suppose there is a topic that should be discussed on both lists (and
actually, this is a real situation, because I am about to start such a
topic in a few days). One way would be to create a new list (call it
usenet-plus-mail) whose only subscribers are ietf-822 and usenet-format.
So if you want to post to both lists, you post instead to
usenet-plus-mail, but you continue posting to the single lists for topics
that only apply to one of them (well, there may be other ways of
arranging things - this is just a possible example).
That is a very uusual special case. In almost all cases,
people do not want to post to sublists except by mistake,
when their intention was to reach the whole list.
8. Loop Control
Is this commonly used? I thought the common idea was that mailing list
expanders should not add Resent- headers. Am I wrong?
Hmmm! I have been told by somebody else that this was not so, but it is
not my reading of DRUMS:
"3.6.6 Resent fields
Resent fields SHOULD be added to any message which is reintroduced by a
user into the transport system. A separate set of resent fields SHOULD be
added if this occurs multiple time.... [and much else]"
A mailing list is not a user. Mailing list expansion does
not mean that a message is taken out of the transport
system and reintroduced.
Surely, that covers mailing list expansion precisely? The message comes
through the transport system to the list server. The list server looks at
it, takes it apart, decides what to do with it, and then puts it back into
the transport system with a new load of recipients, but in a form which
still makes it appear to each ultimate recipient that it came from the
original originator, with the original message-id, and that replies should
go back to him (modulo any Reply-To address). The fact that the "user" is
a bot is neither here nor there. We are used to the idea that "users" on
the internet are often bots.
Looking at what happens in the real world, I see that this list (expanded
by Majordomo, I believe) does NOT add Resent fields, but the usenet-format
list (expanded by Listserv, I believe) DOES. So the real world is clearly
confused on this issue.
If your opinion is the agreed opinion in DRUMS, I will
write so. But very few existing mailing list expanders do
add resent-headers. Are they all wrong? Is not a standard a
description of how software does work in reality?
And if the expander is not supposed to add Resent fields, what is it
supposed to add? It needs something to show it has been involved (and to
detect those loops at the same time).
It should add a Received header, and it should add the
List-headers specified in RFC 2369.
Jacob Palme <jpalme(_at_)dsv(_dot_)su(_dot_)se> (Stockholm University and KTH)
for more info see URL: http://www.dsv.su.se/~jpalme