Harmonzing Internet e-mail and Usenet News protocols
Jacob Palme
+-------------------------------------------+
! Draft user requirements for harmonzing !
! Internet e-mail and Usenet News protocols !
+-------------------------------------------+
Definitions: The word "discussion group" or only "group" is
used below to refer to services for exchanging messages in a
group of people, irrespective of whether this service is
provided by Usenet Newsgroups or by e-mail mailing lists.
Users perceive personal e-mail, distribution lists and Usenet
News as very similar services. They want these three services
to behave in similar ways and they want to be able to use the
same or similar commands in all three environments.
These user requirements of course reflect directly on the
user interface of client software, not on the protocol. But
the e-mail (RFC822 as extended by other RFC-s) and Usenet
News (RFC 1036) protocols can be designed to make it easier
or more difficult to make client software which satisfies the
user requirements below.
In particular, users have the following requirements (all
users do not necessarily have all these requirements):
Client software interactivity:
-----------------------------
1. It should be possible to provide neat, easy-to-use user
interfaces to users. Such interfaces should hide
technical complexity to ordinary users.
Example: Users should not have to format syntactically
correct commands as text in messages they have to send to
mailing list servers. These commands should be
standardized, so that they can be hidden from non-
technical users by good user interfaces.
2. Users who have other languages than English as their
native tongue, should not, unless they so prefer, be
forced to use English-language user interfaces.
Example: Standards for heading fields and network
commands allow the user interface to translate these to
the language of the user.
3. Many users have learnt to use Gopher and WWW clients as
general-purpose client software to access many different
servers. Such users should be able to use this client
software also for performing administration of their
participation in discussion groups.
4. Users should be able to read new messages sorted by
discussion group, and be able to read personal messages,
sent to them individually, separately from messages
coming from mailing lists.
Format of displaying messages:
-----------------------------
5. Similar format for viewing messages in all three
environments (personal e-mail, mailing lists,
newsgroups). This especially applies to message headers.
6. The same heading field should have the same or similar
meaning in all three environments.
Examples: Usenet News should not use "Supersedes:" when
e-mail uses "Obsoletes:" (RFC 1327) and Usenet News
should not use "Expiry Date:" (RFC 1036) when e-mail uses
"Expires:" (RFC 1327).
7. Information given on to whom (individuals, lists and
newsgroups) a message was sent should be consistent in
all three environments.
Example: Information about which groups a message was
sent to should preferably not be labelled "Newsgroups:"
in one environment and "To:" in another environment.
8. Information on where to send replies shoud be consistent
in all three environments.
Examples: "Reply-to:" could indicate where personal
replies (only to the author) should be sent in all three
environments, and "Followup-To:" could indicate where
group replies are to be sent in all three environments.
9. The same name should preferably be shown to the users,
when referring to a discussion group, whether the group
appears as a Usenet Newsgroup or as a mailing list or
both. If this is not technically possible, at least the
names should be easily related to each other in a way
which is understandable to the users.
Handling of conversations:
-------------------------
10. Conversations (chains/trees of messages referring to each
other) should be handled in ways which makes it easy for
users to scan such chains, and to withdraw from a
conversation without unsubscribing from the whole group.
11. It should be possible to let a person who is not a member
of a discussion group participate in a particular
conversation going on within that group.
Distribution of the same contribution to multiple groups:
--------------------------------------------------------
12. Users should be able to easily send the same message, at
the same time, to one or more newsgroups, one or more
distribution lists and one or more individual recipients.
13. Recipients should not be forced to see the same message
more than once even though the message was sent to more
than one newsgroup or distribution list.
E-mail compatibility:
--------------------
14. Users who have e-mail but no other Internet access should
be able to participate in discussion groups with similar
commands, independent of whether the group is a newsgroup
or a distribution list.
15. All Usenet Newsgroups should be available as mailing
lists for those users who only have e-mail connectivity.
(??) Is there a reverse requirement that all mailing
lists should be available also as some kind of Newsgroup?
16. Users of existing services should be able to continue
using as far as possible the services in the way they
have become accustomed to.
Conclusion: The service should interwork well with
existing protocols for e-mail (both Internet and X.400 e-
mail) and for Usenet News.
Administration and archiving:
----------------------------
17. Users should be able to use similar commands in order to
find newsgroups and distribution lists, to subscribe to
them and to unsubscribe to them.
18. Moderated and un-moderated, closed or open discussion
groups should as far as possible be handled with similar
commands for ordinary users. This includes commands to
submit contributions and to ask for membership in the
group.
19. Discussion groups should be either open to anyone, or
open to specified groups (such as employees of a
particular company or members of a particular society),
or closed where all new members must be approved by the
administrator of the group.
20. Users can be ordinary members of discussion groups or
have special roles such as moderators or administrators.
Also moderators and administrators more and more often
are people without technical expertise in the computer
area. Thus, the user interface should be friendly not
only for ordinary users but also for users with special
roles like moderators and administrators.
21. Users should be able to access archives of old messages,
if such exist, with similar commands, irrespective of if
the messages are stored in a Usenet News server or in
some kind of mailing list archive.
In order to satisfy these requirements, there should be
standards for the functions listed below. All the functions
should be available through mail servers, to support users
who only have e-mail access, but the functions should also be
available through direct network connections so that users
with full Internet access can get direct responses to their
commands from the server.
Facilities for ordinary users:
1. Find information about available discussion groups.
2. Subscribe to discussion groups.
3. Ask the administrator of closed groups to let them to
become members.
4. Withdraw from discussion groups.
5. Submit contributions.
6. Submit contributions obsoleting their own previously
submitted contributions.
7. Read contributions.
8. Retrieve and find old contributions.
Facilities for moderators:
9. Find submissions.
10. Accept or reject submissions.
Facilities for administrators:
11. Start new discussion groups.
12. Announce new discussion groups to prospective members.
13. Find requests for membership to discussion groups.
14. Add and remove members from discussion groups.
15. Cancel discussion groups.
16. Appoint moderators and/or new administrators.
17. Modify the attributes of discussion groups. Examples of
such attributes are whether the group is moderated, where
and for how long archives are kept etc.
------------------------------------------------------------------------
Jacob Palme E-mail: jpa...@dsv.su.se
Phone: +46-8-664 77 48 or +46-8-16 16 67
Department of Computer and Fax: +46-8-664 77 48 between 9 am & 2 pm WET
Systems Sciences (DSV) Postal address: Skeppargatan 73,
Stockholm University S-11530 Stockholm, Sweden
John Stanley
Jacob Palme <jpa...@dsv.SU.SE> wrote:
>Users perceive personal e-mail, distribution lists and Usenet
>News as very similar services. They want these three services
>to behave in similar ways and they want to be able to use the
>same or similar commands in all three environments.
I don't accept the assumption. Users who understand the difference
between mail and news (i.e., private and public messages) do not
perceive mail and news to be "very similar".
It is only the people who deal with messaging without considering that
dichotomy that think they are "very similar".
Users that don't understand the difference between public and private
messages need help learning the difference, not a standard that will
help keep them confused.
> Example: Users should not have to format syntactically
> correct commands as text in messages they have to send to
> mailing list servers. These commands should be
> standardized, so that they can be hidden from non-
> technical users by good user interfaces.
This standard implies either 1) everybody scrap every mailing list
server except one, or 2) everybody use a standard for creating mailing
addresses, with a different, orthogonal standard for each mailing list
server. Only by having rule 2 can a mail agent hope to know what
commands to send, based on the address things are sent to.
> Examples: Usenet News should not use "Supersedes:" when
> e-mail uses "Obsoletes:" (RFC 1327) and Usenet News
> should not use "Expiry Date:" (RFC 1036) when e-mail uses
> "Expires:" (RFC 1327).
Expiry date?
>7. Information given on to whom (individuals, lists and
> newsgroups) a message was sent should be consistent in
> all three environments.
Interfaces must show to whom the message will be sent. Not just to whom
it was sent. By then, it is too late.
>11. It should be possible to let a person who is not a member
> of a discussion group participate in a particular
> conversation going on within that group.
Why should group membership not be a prerequisite for participation in
the group? How do you intend that this user who is not a member receive
the replies members send to the group? Telepathy?
>14. Users who have e-mail but no other Internet access should
> be able to participate in discussion groups with similar
> commands, independent of whether the group is a newsgroup
> or a distribution list.
How do you intend this miracle?
>15. All Usenet Newsgroups should be available as mailing
> lists for those users who only have e-mail connectivity.
Ah, the miracle is not clear. Are you volunteering your computer to do
this? How do you force this to happen? And why is it mandatory for
email only users to have USENET connectivity?
>17. Users should be able to use similar commands in order to
> find newsgroups and distribution lists, to subscribe to
> them and to unsubscribe to them.
This requirement is even more limiting that the earlier one. Every
mailing list must have "similar" sub and unsub commands. That implies
everyone runs the same one.
Robert A. Rosenberg
(John Stanley) wrote:
>In article <Pine.SUN.3.90.941016190821.8655A-100000@ester>,
>Jacob Palme <jpa...@dsv.SU.SE> wrote:
>>11. It should be possible to let a person who is not a member
>> of a discussion group participate in a particular
>> conversation going on within that group.
>
>Why should group membership not be a prerequisite for participation in
>the group? How do you intend that this user who is not a member receive
>the replies members send to the group? Telepathy?
>
One reason is to allow the user to send a general "How do I join the list?"
message to the list (which would be replied to via an Email to their address
not to the list address) when they know the list address but not the
subscribe address (which can be listname-request@x [for listname@x],
listserv@x2 [x2 not necessarily equal to the message domain], majordomo@x2
[same caveat], random@x2 [manual subscription], etc).
As to participation in a list, any replies to the user's submission could
have him CC'ed so he gets that thread without getting the rest of the
threads on that list.
John Stanley
>Huh?? I disagree. I know the difference and they still are all just
>messages to be read and possibly answered.
How you answer them should depend on the nature of the reply. E.g., a
private reply to someone asking a question should probably contain the
answer. A public reply to a common question may only contain a pointer
to the answer. (The latter saves bandwidth and is less annoying to those
reading the group who already know the answer, while letting those who
don't know look it up for themselves.)
>What dichotomy?
Keep up with the discussion. Public vs. private.
>>Users that don't understand the difference between public and private
Oh, look, I answered your question in the next sentence already.
>There is really only one difference that should be significant at the
>user's response level. It is appropriate to respond back to the group
>that originally saw a message and quote the original in the body. It
It may or may not be. For example, someone mentioned sending "how do I
subscribe" messages to a group. It would be completely INappropriate to
reply to these messages in the group. (Those who read the group already
know how to subscribe or they wouldn't be reading it.)
>is also appropriate and common to respond privately to the sender. It
So? They type of reply, and what it says, will often depend on the
nature of the message being replied to. To blur the difference between
public and private messages is to open the gate for people to think they
are replying privately when posting, or publicly when mailing privately.
>is a judgement call as to whether or not it is appropriate to add new
>recipients or groups to the response when the original had more than
>one recipient.
This is not the issue.
>However, it is never appropriate to add new recipients
>to a response to a private message without the original sender's
>permission. This relates to the sender's rights and expectations rather
>than the transport mechanism, though.
And making it all look the same removes the perceived limitation that
you say is never appropriate.
>>Interfaces must show to whom the message will be sent. Not just to whom
>>it was sent. By then, it is too late.
>
>Some do, some don't unless you go out of your way to look.
Then those that don't should be fixed. They are broken. I would not
allow a mail agent that did not report prior to sending mail the
To: address (and any and all CC: and BCC:) to exist on my systems.
It would be a disservice to my users.
>But the
>problem that relates to my comment above is that it is seldom possible
>to tell exactly who has seen (will see) the original.
This capability is well near impossible, especially for news. And I
certainly do not want to see, nor would I burden my mailing list readers
with, a complete list of who a mailing list message was sent to.
>Another problem is that some people (or their software) typically
>reply to a news posting with simultaneous email to the poster and
>a posting of their own. I often see the email reply first and respond
>privately, then later see the posted response and realize that the
>discussion should have continued in the newsgroup.
This is a problem that user education will solve. Rudeness such as this
is not a property of the software, and will not stop just because a new
header has been created.
>What constitutes "group membership"? A set of CC:'s in an email header
>should be enough to allow email replies to everyone.
Unless, of course, the group does not support CC: headers at all.
>Many, but perhaps
>not all mailers have commands to reply to everyone mentioned in the
>headers.
And many allow you to reply only to specific persons. If a question is
asked to a list, and it is appropriate to answer in the list, I will do
that, with the assumption that anyone who is wanting information from people
will make a tiny effort to go to where the people are, not demand that
they come to him.
>>This requirement is even more limiting that the earlier one. Every
>>mailing list must have "similar" sub and unsub commands. That implies
>>everyone runs the same one.
>
>Are you saying that you enjoy the diversity?
As a user, I neither enjoy nor dislike the diversity. I accept it as the
way things are, and make the effort to find out what the correct access
method is. In case you haven't noticed, almost all of the listings of
mailing lists include the request address. The effort of continuing to
read the paragraph is minimal.
As an admin, I reject the notion that I must waste my time installing a
new mailing list server just because people are too stupid to read the
instructions on how to use the one I have now.
Leslie Mikesell
John Stanley <sta...@skyking.OCE.ORST.EDU> wrote:
>>There is really only one difference that should be significant at the
>>user's response level. It is appropriate to respond back to the group
>>that originally saw a message and quote the original in the body. It
>It may or may not be. For example, someone mentioned sending "how do I
>subscribe" messages to a group. It would be completely INappropriate to
>reply to these messages in the group. (Those who read the group already
>know how to subscribe or they wouldn't be reading it.)
It's inappropriate to send such a message to a group in the first place.
>>is also appropriate and common to respond privately to the sender. It
>
>So? They type of reply, and what it says, will often depend on the
>nature of the message being replied to. To blur the difference between
>public and private messages is to open the gate for people to think they
>are replying privately when posting, or publicly when mailing privately.
Sorry, but I don't blindly assume that everything in email is private,
so there is nothing blurring any distinction here.
>>is a judgement call as to whether or not it is appropriate to add new
>>recipients or groups to the response when the original had more than
>>one recipient.
>
>This is not the issue.
It is exactly the issue. Nothing other than my own judgement can or
should determine the type of reply I send.
>>However, it is never appropriate to add new recipients
>>to a response to a private message without the original sender's
>>permission. This relates to the sender's rights and expectations rather
>>than the transport mechanism, though.
>
>And making it all look the same removes the perceived limitation that
>you say is never appropriate.
On the contrary, there is no limitation, only a choice. Making it look
the same emphasizes the fact that you always have the same choice and
you should make it appropriately.
>>>Interfaces must show to whom the message will be sent. Not just to whom
>>>it was sent. By then, it is too late.
>>
>>Some do, some don't unless you go out of your way to look.
>
>Then those that don't should be fixed. They are broken. I would not
>allow a mail agent that did not report prior to sending mail the
>To: address (and any and all CC: and BCC:) to exist on my systems.
>It would be a disservice to my users.
Ummm, the user should *direct* the choice of replying to only the sender
or to all the recipients. If they can't remember which they chose, it
probably won't help to print the list again.
>>Another problem is that some people (or their software) typically
>>reply to a news posting with simultaneous email to the poster and
>>a posting of their own. I often see the email reply first and respond
>>privately, then later see the posted response and realize that the
>>discussion should have continued in the newsgroup.
>
>This is a problem that user education will solve. Rudeness such as this
>is not a property of the software, and will not stop just because a new
>header has been created.
It's not really rude, since it does call your attention to the response
which might otherwise get lost since no one can possibly read all
the newsgroups every day. It would just be nice if there were a
standard way of indicating in the headers that the message was also
posted to the newsgroup.
>>What constitutes "group membership"? A set of CC:'s in an email header
>>should be enough to allow email replies to everyone.
>
>Unless, of course, the group does not support CC: headers at all.
I meant that a set of CC:'s will create a group by themselves, without
any list processor. But, if you are responding to a list message with
an email interface, your own mailer will process the CC: address for
your reply, although it may be stripped by the mailing list processing
and prevent later responses from getting back to the added address.
Les Mikesell
l...@mcs.com
Mark Crispin
The perceived models of Internet email, as established in common email
software, and of USENET news, as established in common news software, are
fundamentally different. There is no way for these two models to integrate
without creating an integrated model that has differences from the two. There
are different protocols to transport mail and news.
So far, he's correct.
But, from this point, he makes this conclusion:
This state of affairs is right and desirable. Users must be aware that there
are two protocols and two incompatible worlds at all times. Software must not
smooth over these differences, and no attempt made to make these worlds
compatible.
It is here where many of us disagree.
I see absolutely no reason why I should have to care whether something is news
or mail. It's up to these expensive computers to figure that detail out.
Ideally, there would be no Newsgroups: line. Newsgroups would appear on the
To:, cc:, or bcc: lines where they belong. But we live in an imperfect world.
The actual integration effort is much more modest in its scope and much less
disruptive.
Leslie Mikesell
John Stanley <sta...@skyking.OCE.ORST.EDU> wrote:
>>Users perceive personal e-mail, distribution lists and Usenet
>>News as very similar services. They want these three services
>>to behave in similar ways and they want to be able to use the
>>same or similar commands in all three environments.
>I don't accept the assumption. Users who understand the difference
>between mail and news (i.e., private and public messages) do not
>perceive mail and news to be "very similar".
Huh?? I disagree. I know the difference and they still are all just
messages to be read and possibly answered.
>It is only the people who deal with messaging without considering that
>dichotomy that think they are "very similar".
What dichotomy?
>Users that don't understand the difference between public and private
>messages need help learning the difference, not a standard that will
>help keep them confused.
There is really only one difference that should be significant at the
user's response level. It is appropriate to respond back to the group
that originally saw a message and quote the original in the body. It
is also appropriate and common to respond privately to the sender. It
is a judgement call as to whether or not it is appropriate to add new
recipients or groups to the response when the original had more than
one recipient. However, it is never appropriate to add new recipients
to a response to a private message without the original sender's
permission. This relates to the sender's rights and expectations rather
than the transport mechanism, though.
>>7. Information given on to whom (individuals, lists and
>> newsgroups) a message was sent should be consistent in
>> all three environments.
>
>Interfaces must show to whom the message will be sent. Not just to whom
>it was sent. By then, it is too late.
Some do, some don't unless you go out of your way to look. But the
problem that relates to my comment above is that it is seldom possible
to tell exactly who has seen (will see) the original. For one thing
email headers are not required to reflect the envelope addresses,
nor does any single address have to map to any single entity.
Another problem is that some people (or their software) typically
reply to a news posting with simultaneous email to the poster and
a posting of their own. I often see the email reply first and respond
privately, then later see the posted response and realize that the
discussion should have continued in the newsgroup.
>>11. It should be possible to let a person who is not a member
>> of a discussion group participate in a particular
>> conversation going on within that group.
>
>Why should group membership not be a prerequisite for participation in
>the group? How do you intend that this user who is not a member receive
>the replies members send to the group? Telepathy?
What constitutes "group membership"? A set of CC:'s in an email header
should be enough to allow email replies to everyone. Many, but perhaps
not all mailers have commands to reply to everyone mentioned in the
headers. Using these commands to reply to a mailing list message will
send a copy back to the sender whether he is on the list or not (if he
is, he'll get two copies) and add him to the CC: list for subsequent
replies.
>>17. Users should be able to use similar commands in order to
>> find newsgroups and distribution lists, to subscribe to
>> them and to unsubscribe to them.
>
>This requirement is even more limiting that the earlier one. Every
>mailing list must have "similar" sub and unsub commands. That implies
>everyone runs the same one.
Are you saying that you enjoy the diversity?
Les Mikesell
l...@mcs.com
Leslie Mikesell
John Stanley <sta...@skyking.OCE.ORST.EDU> wrote:
>>Sorry, but I don't blindly assume that everything in email is private,
>>so there is nothing blurring any distinction here.
>
>I didn't say everything in mail is private. I am talking about the
>distinction between public and private. News is always public. Mail is
>a mix. But everyone here seems to be dealing with mail only as public.
>It isn't.
But your argument against combining mail and news under a single interface
seem to revolve around the assumption that email is automatically private.
>Making it look the same removes information you need to know in order to
>choose. It may emphasize that you need to make the choice, but it takes
>away your ability to do so.
Are you saying your mail interface does not offer a choice of replying
only to the sender or to all recipients? If that is the case I can
see why you wouldn't want to handle news with it. But, I wouldn't
want it for mail either.
>>Ummm, the user should *direct* the choice of replying to only the sender
>>or to all the recipients. If they can't remember which they chose, it
>>probably won't help to print the list again.
>
>Right. Seeing a list of 80 names in a CC: header won't remind anyone
>that they are replying in an essentially public manner. Seeing that they
>are sending a response to "foobar-list" won't remind them that they
>selected a public response. You must have some really dense users.
No, the opposite. They make the selection. Why should they forget
and need to be reminded? Actually I'm not arguing against showing
the headers - I prefer it but don't consider it to be essential that
you see them at the same time as the message body.
>There is no guarantee that some helpful mailer will not twist the
>presence of a "Newsgroups" header in mail into "I must post this mail
>for the user."
The real problems come from the fact that there is not a standard
way to gateway between mailing lists and news transports or for
a single user interface to handle both transports. But the fact
that current attempts are inconsistent does not mean it is a bad
idea.
Les Mikesell
l...@mcs.com
Mark Crispin
> Email is generally a private communication medium.
This is not supported by measurements at any site that I am familiar with.
The overwhelming amount of mail is mailing-list and *not* person-to-person.
> You have shown a remarkable inability to interpret simple english.
I understand your arguments perfectly well. I ignore arguments which are
patently false and absurd. You just proved it by using ``eh, you no speaka ze
English?'' as a prop. Correct arguments do not need rotten props to shore
them up.
The public/private message dichotomy is well understood. But that dichotomy
is not the SMTP/NNTP dichotomy. It is the reply sender/all dichotomy. Pine
has always made this distinction. Furthermore, it is not possible to post
news in Pine without replying Yes to a ``Posted message may go to thousands of
readers. Really post?'' warning. Both of these facts have been stated before
in this discussion.
According to your headers, you use an old version of Pine (3.87) which does
not have the capability in question. Have you bothered to investigate for
yourself how that capability works in 3.91? Please do so before making
further comments.
James C Deikun
John Stanley <sta...@skyking.OCE.ORST.EDU> wrote:
>>Sorry, but I don't blindly assume that everything in email is private,
>>so there is nothing blurring any distinction here.
>
>I didn't say everything in mail is private. I am talking about the
>distinction between public and private. News is always public. Mail is
>a mix. But everyone here seems to be dealing with mail only as public.
>It isn't.
You make a good point, but it's not in your favor. Mail already
handles both public and private messages. And I see more mailreaders
that read news than newsreaders that read mail. If private and public
messages are already both supported, what of any importance is this
unification supposed to be blurring?
>>It is exactly the issue. Nothing other than my own judgement can or
>>should determine the type of reply I send.
>
>You must base that judgement on facts at hand. If the fact that you are
>replying in a public posting is lost, you may make the wrong decision.
>Similarly, if the fact that you are replying in private is lost, then
>you may decide incorrectly again.
Mechanisms to reply publicly and privately already exist in both news
and mail reading software. I still am not quite sure how using the
same mechanism of making the same distinction with two different
underlying protocols is supposedly going to confuse the hell out of
users.
(snip)
>>Making it look
>>the same emphasizes the fact that you always have the same choice and
>>you should make it appropriately.
>
>Making it look the same removes information you need to know in order to
>choose. It may emphasize that you need to make the choice, but it takes
>away your ability to do so.
Public/private is *not* the same as news/mail, a fact you seem unable
to accept. How do mailing lists (which look just the same as private
mail) take away your ability to choose whether to respond publicly or
privately? Why should unification with news make that any different?
Is there a reason you believe that all unified mail/news UIs will be
done in the worst way humanly possible?
What removes information essential to me is getting mail in response
to a news discussion and being unable to determine if it was also
publicly posted because of the broken newsreaders that leave spurious
Newsgroups: headers in purely private mail.
(snip)
>>It would just be nice if there were a
>>standard way of indicating in the headers that the message was also
>>posted to the newsgroup.
>
>As long as that header is NOT a standard news header, I have absolutely
>NO PROBLEM with this. I have even said as much, in public. I have even
>gone as far as suggesting names for the new mail headers.
>
>The problem I have is when people want news headers to play double duty.
>There is no guarantee that some helpful mailer will not twist the
>presence of a "Newsgroups" header in mail into "I must post this mail
>for the user." In fact, I can almost guarantee that it WILL happen.
>I have found at least one "helpful" mailer that made just such a
>decision based on the USER NAME the mail was addressed to, completely
>ignoring the tiny detail that the mail was addressed to a machine in a
>completely different domain. If you start using a header that some
>mailers leave in the mail they send in reply to news articles, you are
>guaranteeing that some helpful soul will program his mail system to obey
>it when it was not the intention of the user for it to be obeyed, nor
>his choice to include it.
Mailers that leave Newsgroups: headers in mail in reply to news articles
are broken. Period. When news and mail started using one header
format, and mail replies to news were first supported, it immediately
became a matter of common sense (that a lot of people apparently
failed to catch on to) to keep the possible sets of headers for mail
and news, if not identical, at least mutually consistent. With the
first mail<->news gateways this became even more painfully obvious.
The current behavior of certain newsreaders is wrong, has always been
wrong, and ought to be fixed.
As for "helpful" mailers, mailers aren't any more confused by the
current practice than are a lot of users.
(remainder omitted)
--
James Deikun, University of Pittsburgh
#include <std_disclaimer.h>
door (n.): An object that a cat is always on the wrong side of.
John Stanley
>> Email is generally a private communication medium.
>
>This is not supported by measurements at any site that I am familiar with.
>The overwhelming amount of mail is mailing-list and *not* person-to-person.
I am tired of this showing up in my mailbox. My mailbox is for matters
I need to deal with quickly. I do not intend on continuing this
discussion by mail.
If you feel it is so important that you cannot wait the day or two it
might take me to see your witty rebuttals in news, that is your
problem. Do not make it mine.
John Stanley
Robert A. Rosenberg <hal...@panix.com> wrote:
>In Article <381jbb$4...@gaia.ucs.orst.edu>, sta...@skyking.OCE.ORST.EDU
>(John Stanley) wrote:
>>Why should group membership not be a prerequisite for participation in
>>the group? How do you intend that this user who is not a member receive
>>the replies members send to the group? Telepathy?
>>
>One reason is to allow the user to send a general "How do I join the list?"
>message to the list (which would be replied to via an Email to their address
>not to the list address) when they know the list address but not the
>subscribe address (which can be listname-request@x [for listname@x],
1. This type of question is not participation, it is asking how to
participate. It is grossly off-topic for any mailing list or newsgroup,
except perhaps for the "how-do-I-subscribe-to-mailing-lists" mailing
list.
2. You do NOT want to encourage sending this sort of request to the
list EVER. For a 1000 member list, you will be bothering 1000 people to
answer a question that half have probably forgotten, and that one
should do: the list maintainer. For a USENET group, you are wasting an
incredible anmount of net resources for something that most people
couldn't answer anyway (it depends very much on what system with what
software is being used.)
If you can't make the effort to find out for yourself how to subscribe,
then your need to subscribe isn't very great, and is certainly less
than your right to annoy people who want to read about a certain topic
with "how do I join" messages.
>As to participation in a list, any replies to the user's submission could
>have him CC'ed
That is, if the submission to the list were marked in some way to
indicate this was necessary, if it could even be marked that way. There
is no header in news that will indicate this function.
And you need to read USENET for a while to notice the general opinion
that "please email me an answer to this question" is rude and a waste of
the net's time. If the question isn't important enough to make the
effort to participate in the group, it isn't important enough to warrant
an answer, and asking for personal, private replies from people who are
taking their time to give you information is rude to the rest of the
users who might benefit from the answer, too.
Mark Crispin
> I am tired of this showing up in my mailbox. My mailbox is for matters
> I need to deal with quickly. I do not intend on continuing this
> discussion by mail.
>
> If you feel it is so important that you cannot wait the day or two it
> might take me to see your witty rebuttals in news, that is your
> problem. Do not make it mine.
Foo. Nobody is impressed by ``my mailbox is too important'' whines.
If you're upset about replies going to your bubbles.wes.army.mail mailbox
instead of your skyking.oce.orst.edu mailbox, you should have thought about
that when you first sent mail from bubbles on October 21.
You have posted several messages to HEADER-PEOPLE, an email mailing list which
has been around for over 17 years. Not only that, but after sending the above
message you proceeded to post two additional messages to HEADER-PEOPLE, 22 and
33 minutes later, as attached below.
Mailing lists are a two-way street. If your mailbox is too important to
receive email replies, don't send email in the first place. You have already
stated your position. Saying the same thing over and over again isn't going
to change the minds of those who disagree. If you don't have anything new to
say, then drop out of the discussion; you won't get any more responses in your
Very Important Mailbox afterwards.
Date: 23 Oct 1994 14:27:02 GMT
From: John Stanley <sta...@skyking.oce.orst.edu>
Subject: Re: News vs. Email vs. News vs...
To: header...@mc.lcs.mit.edu
In article <1994102115...@prodigy.bc.ca>,
Brian Edmonds <bedm...@prodigy.bc.ca> wrote:
>Actually, if I encrypt a message and dump it onto a newsgroup, I would
>consider that to be more private than a plaintext email message
And I can package up news and distribute via mail. That doesn't make any
other piece of email I send any less private.
Date: 23 Oct 1994 14:38:19 GMT
From: John Stanley <sta...@skyking.oce.orst.edu>
Subject: Re: Harmonzing Internet e-mail and Usenet News protocols
To: header...@mc.lcs.mit.edu
In article <4ie34RiSM...@transarc.com>,
<Craig_E...@transarc.COM> wrote:
>Where I live, most mail is in fact
>public--shared among people in named groups.
Where I work, most mail I generate is private. Most mail I get is
private, too.
Where I live, almost all of it is private.
>My advice would be to abandon the mail/news distinction, and instead
>push MUA developers to ensuring that users know the difference between
>reply-to-sender and reply-to-all/reply-to-recipients.
And if the "sender" is a mailing list or USENET newsgroup? People who
think "reply to sender" is a private message and thus dump their life
histories into it, only to find that "sender" was a mail-news gateway
are going to be rather unhappy. "Hey, that was supposed to be private! I
used the private reply funtion!" "No, you used 'reply-to-sender'. There
isn't any 'private reply' function."
>on the To: field. Given that the prefix is ``netnews.'', it's unlikely
>to be part of a legitimate user's name. Should that situation arise,
>the MUA will query the message-composer as to which recipient was
>intended.
The recipient may very well have no choice. If a "helpful" mail
transport decides that it should post the message instead of passing it
to the system in the To: address, because it is more efficient, what
matters if the sender intended it to be a private message?
>> It is not acceptable to me.
>
>What would you propose, that doesn't presume that all mail is private?
Create a new set of headers, that are NOT already defined to mean
something else. For example: Also-mailed-to: and Also-posted-to:.
Of course, I suggested these already.
John Stanley
On Mon, 24 Oct 1994, Mark Crispin wrote:
> Foo. Nobody is impressed by ``my mailbox is too important'' whines.
Foo yourself. I don't give a shit if you are impressed or not. I have
asked you not to dump this stuff in my mailbox, and to let it wait until
I have time to deal with it in a proper manner -- that is, when I get
time to read news and have time to reply to your arguments.
You have continued to ignore that request. Why, I don't know. Maybe you
don't understand simple english.
You have chosen to participate in mail. That is your right. I have chosen
to participate in news. That is my right. Now, stop mailing me this crap.
Is that clear enough for you?
> Mailing lists are a two-way street. If your mailbox is too important to
> receive email replies, don't send email in the first place.
You sent me the first email of this exchange. I made the mistake of
replying to it. I tried to rectify that mistake, but you continue to
ignore a simple request to remove my name from your mailings. Why?
Rahul Dhesi
>My advice would be to abandon the mail/news distinction, and instead
>push MUA developers to ensuring that users know the difference between
>reply-to-sender and reply-to-all/reply-to-recipients.
The trouble with this discussion is that there are sharp differences
between mailing lists and Usenet newsgroups. ('Vive la Overlap' does
not imply 'Le plus des changes, le plus des meme-choses', if you will
pardon my French.) In the context of a Usenet newsgroup 'reply to
sender' has some meaning (interpretation: send private reply to author
of posting). But 'reply to recipients' has no meaning in this
context.
The correct choice, when discussing email vs News, is between 'send
private reply to author' and 'post public follow-up'. A follow-up is
never sent to any specific people. A follow-up, from the point of view
of the person posting it, has no specific recipients. Anybody who
chooses to read that follow-up -- and there is no predefined list of
such people -- becomes a recipient. Usenet is far more dynamic than
may appear at first sight.
You might call this a trichotomy:
1. Send private response to original author.
2. Send public response to all recipients, if you can identify them.
3. Send public follow-up to various news spool directories, where
countless unidentifiable people will have access to it.
--
Rahul Dhesi <dh...@rahul.net>
Leslie Mikesell
John Stanley <sta...@skyking.OCE.ORST.EDU> wrote:
>>Where I live, most mail is in fact
>>public--shared among people in named groups.
>Where I work, most mail I generate is private. Most mail I get is
>private, too.
Just out of curiosity, do you not have groups of people with common
interests where you work or do they have a better way than email to
communicate?
Les Mikesell
l...@mcs.com
John Stanley
James C Deikun <jcds...@pitt.edu> wrote:
>If private and public
>messages are already both supported, what of any importance is this
>unification supposed to be blurring?
If the nose of the camel is already in the tent, why worry about the
rest of him coming in?
>Mechanisms to reply publicly and privately already exist in both news
>and mail reading software. I still am not quite sure how using the
>same mechanism of making the same distinction with two different
>underlying protocols is supposedly going to confuse the hell out of
>users.
Users are already confused enough. You can't imagine how they will get
more confused when news comes to them through the same program that
private messages do?
>Public/private is *not* the same as news/mail, a fact you seem unable
>to accept.
For many people, that is the distinction, a fact that all the experts on
every system they know about don't want to accept.
>Mailers that leave Newsgroups: headers in mail in reply to news articles
>are broken. Period.
Yes, I have already said that. They exist, however, and ignoring them
and doing the wrong thing with someone's mail because their admin
installed a broken newsreader is not the correct answer.
>When news and mail started using one header
>format, and mail replies to news were first supported, it immediately
>became a matter of common sense (that a lot of people apparently
>failed to catch on to) to keep the possible sets of headers for mail
>and news, if not identical, at least mutually consistent.
They are.
>The current behavior of certain newsreaders is wrong, has always been
>wrong, and ought to be fixed.
Yes. Your point?
>As for "helpful" mailers, mailers aren't any more confused by the
>current practice than are a lot of users.
One current practice is to name the mailbox for a mail-news gateway
something like comp.mai...@some.site.com. I have had a helpful
mailer take mail addressed to such a username and post it to the
newsgroup. The only problem with this is that the hostname part of the
address was in a completely different domain. This is a violation of the
RFC's for mail handling. It did it anyway, however.
I have no doubt that, should "Newsgroups:" be defined in mail to mean
"post this to these groups, too", someone will program their mailer to
process this header even if they aren't the site the mail is being sent
to. Since creating this new header for mail will not automatically
change the known behaviour of existing newsreaders, it is almost
guaranteed that people who think they are making private replies will
wind up with their mail posted to news.
Of course, creating new headers that contain the information you want
(and, for the slow readers, changing software to honor and create those
headers) will not have such a problem.
John Stanley
Dave Crocker <dcro...@mordor.stanford.EDU> wrote:
>Many newsgroups are contained within special bounds, such as a company.
>These are very much NOT public.
Considering the number of public newsgroups and the number of private
newsgroups any company is likely to have, public vastly outweighs the
private.
Rahul Dhesi
(Mark Crispin) writes:
>You have posted several messages to HEADER-PEOPLE, an email mailing list which
>has been around for over 17 years. Not only that, but after sending the above
>message you proceeded to post two additional messages to HEADER-PEOPLE, 22 and
>33 minutes later, as attached below.
>Mailing lists are a two-way street....
(So are email/Usenet gateways, wherein lies the problem.)
It's a pity that the 'header-people' see everything in email. I believe
this is why so many of them do not see the Usenet users' vastly
different perspective. Usenet is very different from mailing lists.
Gatewaying Usenet to a mailing list is like gatewaying an elephant to
its legs.
(Visualize the famous elephant picture, with different people feeling
different parts of the beast. The header-people see only the legs.)
--
Rahul Dhesi <dh...@rahul.net>
John Stanley
<Craig_E...@transarc.COM> wrote:
>Excerpts from internet.header-people: 23-Oct-94 Re: Harmonzing Internet
>e-m.. John Sta...@skyking.oce (1654)
>
>> >My advice would be to abandon the mail/news distinction, and instead
>> >push MUA developers to ensuring that users know the difference between
>> >reply-to-sender and reply-to-all/reply-to-recipients.
>
>A truly broken gateway.
No, a properly working gateway. Gateways that claim that I mailed a
message when I did not are broken. Gateways that don't tell you that
mail came from them are hiding the fact that the mail came from them.
>By the way, ``sender'' wasn't meant to mean
>contents of the Sender: field.
That is why "sender" is in quotes.
>> The recipient may very well have no choice. If a "helpful" mail
>> transport decides that it should post the message instead of passing it
>> to the system in the To: address, because it is more efficient, what
>> matters if the sender intended it to be a private message?
>
>It's the sender who has the choice. I don't understand this at all.
That's right. I meant to write "the sender may very well have no
choice."
>If the mail transport is posting rather than mailing, it's broken.
That's right, but it happened. But when there becomes a "standard" that
says that a "newsgroups" header means mail is supposed to be (or
already was) posted, you will be hard pressed to say that that same
transport is broken. "That header says it will be posted, so what's the
problem if I do it here?"
>What problem are we trying to solve? Coding around a broken transport
>layer that posts rather than mails?
No. Preventing things from breaking when someone creates a new meaning
for existing headers.
Dave Crocker
>News is always a public medium.
News is always for GROUP communications, but there is a very big difference
between having a conversation in a group and having one in public. A
product planning meeting is group-based, but certainly isn't public.
Many newsgroups are contained within special bounds, such as a company.
These are very much NOT public.
d/
--------------------
Dave Crocker
Brandenburg Consulting Phone: +1 408 246 8253
675 Spruce Dr. Fax: +1 408 249 6205
Sunnyvale, CA 94086 Email: dcro...@mordor.stanford.edu
John Gardiner Myers
> of posting). But 'reply to recipients' has no meaning in this
> context.
>
> The correct choice, when discussing email vs News, is between 'send
> private reply to author' and 'post public follow-up'. A follow-up is
> never sent to any specific people. A follow-up, from the point of view
> of the person posting it, has no specific recipients. Anybody who
> chooses to read that follow-up -- and there is no predefined list of
> such people -- becomes a recipient.
This is really no difference than with mailing lists--a
reply/follow-up to a mailing list is not necessarily to any predefined
list of recipients. In many cases, anybody who chooses to read the
reply/follow-up can pick it up from a list archive.
Rahul Dhesi <dh...@rahul.net> writes:
> In <78267823...@nifty.andrew.cmu.edu> Chris Newman
> <chr...@CMU.EDU> writes:
>
> >At CMU, we store mailing lists in publicly readable (non-netnews)
> >mailstores. Therefore this is not a fundamental difference between
> >the objects -- merely a common implementation difference between
> >mailing lists and usenet.
>
> A better description of your implementation is needed.
The address "arpalists+h...@andrew.cmu.edu" is subscribed to
the header-people mailing list. Any mail sent to this address gets
delivered to the folder named "internet.header-people", where it can
be read by anyone on the system.
sta...@skyking.OCE.ORST.EDU (John Stanley) writes:
> Users are already confused enough. You can't imagine how they will get
> more confused when news comes to them through the same program that
> private messages do?
As I've mentioned, andrew.cmu.edu has been doing this for about eight
years now, with no significant confustion problems. The
public/private distinction is made through the namespace, not through
the program.
Rahul Dhesi <dh...@rahul.net> writes:
> It's a pity that the 'header-people' see everything in email. I believe
> this is why so many of them do not see the Usenet users' vastly
> different perspective. Usenet is very different from mailing lists.
> Gatewaying Usenet to a mailing list is like gatewaying an elephant to
> its legs.
>
> (Visualize the famous elephant picture, with different people feeling
> different parts of the beast. The header-people see only the legs.)
Having worked with an integrated news/bboard/mail system for several
years, I visualize the "news is fundamentally different" folks like
this. Most Usenet users have used vastly different interfaces for
accessing news and mail, they think they must therefore be
fundamentally different. It's like they only see two legs, one marked
"mail", the other "news". Look up.
--
_.John G. Myers Internet: jg...@CMU.EDU
LoseNet: ...!seismo!ihnp4!wiscvm.wisc.edu!give!up
Keith Moore
> Users are already confused enough. You can't imagine how they will get
> more confused when news comes to them through the same program that
> private messages do?
But this problem (public versus private traffic) already exists, and needs to be
solved, in a pure-email world. While we're at it it may make sense to address
email-news integration at the same time.
Seems like the thing to do is to explicitly distinguish "public" traffic from
"private" traffic in the message format, regardless of what transport is being
used. For news this extra header would always appear; for email, it would only
appear on lists.
Of course this solution wouldn't help matters in the short term. But there
doesn't appear to be a short-term solution anyway.
--
Keith Moore NETLIB development group
Computer Science Department / University of Tennessee at Knoxville
107 Ayres Hall / Knoxville TN 37996-1301
Let's stamp out encryption export laws in our lifetime.
James C Deikun
John Stanley <sta...@skyking.OCE.ORST.EDU> wrote:
>In article <38cvpm$7...@usenet.srv.cis.pitt.edu>,
>James C Deikun <jcds...@pitt.edu> wrote:
>>If private and public
>>messages are already both supported, what of any importance is this
>>unification supposed to be blurring?
>
>If the nose of the camel is already in the tent, why worry about the
>rest of him coming in?
Mailing lists and mail<->news gateways are not the "nose" of the camel.
The "whole camel" is already happily entrenched in the "tent." There
is no difference in principle between these things and actually reading
mail and news with the same reader.
>>Mechanisms to reply publicly and privately already exist in both news
>>and mail reading software. I still am not quite sure how using the
>>same mechanism of making the same distinction with two different
>>underlying protocols is supposedly going to confuse the hell out of
>>users.
>
>Users are already confused enough. You can't imagine how they will get
>more confused when news comes to them through the same program that
>private messages do?
For any user who gets mailing lists, or any user whose sysadmin makes
announcements of any kind via mail, public messages have already come
to them through the same program that private messages do. For anyone
who's used a BBS, this is likewise true. It's not as if this is a
revolutionary new concept.
>>Public/private is *not* the same as news/mail, a fact you seem unable
>>to accept.
>
>For many people, that is the distinction, a fact that all the experts on
>every system they know about don't want to accept.
I submit that this is not going to cause any more confusion than the
ability to send private mail through rn already does. News and mail
are already partly integrated in the same interface, yet the user
still has to learn to use two different programs, and, if he or she
can't figure out how to set his or her editor, often ends up using
two COMPLETELY different interfaces to perform the same function.
I don't understand what kind of reasoning leads you to proclaim that
this is LESS confusing than unification.
>>Mailers that leave Newsgroups: headers in mail in reply to news articles
>>are broken. Period.
>
>Yes, I have already said that. They exist, however, and ignoring them
>and doing the wrong thing with someone's mail because their admin
>installed a broken newsreader is not the correct answer.
A good kludge for now would be to zap the Newsgroups: header or modify
it to something on the order of Part-of-a-discussion-originally-on:
(although I'm not sure what the use of the latter would be) if the
unified mail/news reader cannot verify the existence of the message
as news. This would be done while importing messages to local storage,
which Pine (among other mailers) does.
This of course would be wasteful and not 100% accurate, but it would
be wrong less of the time, could be dispensed with if the great
majority of all newsreaders ever end up fixed, and might even help to
bring an earlier end to this thread. :-)
>>When news and mail started using one header
>>format, and mail replies to news were first supported, it immediately
>>became a matter of common sense (that a lot of people apparently
>>failed to catch on to) to keep the possible sets of headers for mail
>>and news, if not identical, at least mutually consistent.
>
>They are.
The presence of a Newsgroups: header in a news article means the
article was posted to the indicated newsgroups. Thanks to bad
newsreaders, and as you repeatedly emphasize, in current practice
the presence of a Newsgroups: header in mail currently means no
such thing. This is obviously an inconsistency, if you use any
sane brand of logic, and ought to be fixed.
>>The current behavior of certain newsreaders is wrong, has always been
>>wrong, and ought to be fixed.
>
>Yes. Your point?
If you don't get it by now, I have my doubts you ever will.
>>As for "helpful" mailers, mailers aren't any more confused by the
>>current practice than are a lot of users.
>
>One current practice is to name the mailbox for a mail-news gateway
>something like comp.mai...@some.site.com. I have had a helpful
>mailer take mail addressed to such a username and post it to the
>newsgroup. The only problem with this is that the hostname part of the
>address was in a completely different domain. This is a violation of the
>RFC's for mail handling. It did it anyway, however.
This is why I would like to see that role belonging to Newsgroups:
instead of To:. One particular broken mailer does not necessarily
indicate a fundamental problem with the endeavour of unifying the
interface to mail and news, any more than the existence of MS-DOS
shows the futility of providing computers with operating systems.
>I have no doubt that, should "Newsgroups:" be defined in mail to mean
>"post this to these groups, too", someone will program their mailer to
>process this header even if they aren't the site the mail is being sent
>to. Since creating this new header for mail will not automatically
>change the known behaviour of existing newsreaders, it is almost
>guaranteed that people who think they are making private replies will
>wind up with their mail posted to news.
Since the Newsgroups: header would be a MESSAGE header rather than
specifically a mail header, and since it's not used by SMTP, and
simply because of the semantics of it, the sender/posters site not
only would be justified in handling the header, but would be
unjustified in NOT handling the header if it has performed some
sanity checks to make sure it is not including such headers as
the result of someone's broken newsreader screwing up.
>Of course, creating new headers that contain the information you want
>(and, for the slow readers, changing software to honor and create those
>headers) will not have such a problem.
It would be preferable in creating a truly unified interface to make
sure that at least the *internal* representations of mail and news
are self-consistent and unified. Handling brokenness is the
prerogative of a preprocessing step.
--
James Deikun, University of Pittsburgh
#include <std_disclaimer.h>
It would be so nice to get the newspaper in one's mailbox.
Peter da Silva
John Gardiner Myers <jg...@CMU.EDU> wrote:
> As I've mentioned, andrew.cmu.edu has been doing this for about eight
> years now, with no significant confustion problems. The
> public/private distinction is made through the namespace, not through
> the program.
Exactly. There's a significant difference between a broadcast and a multicast
list, but that can be maintained independantly of the header. Both the
statement that unification is undesirable and the statement that the two
mechanisms are basically the same are nonsense.
Broadcast entities, particularly distributed broadcast entities, are open
in a way that even systems like fidonet echoes don't see... and don't have
to deal with. Can you imagine someone posting
I AM THE TERROR THAT FLAPS
NO CARRIER
multiple times to multiple fido echoes, or mailing lists, or moderated groups
for very long before the offending site was cut out of the feed? Can you
imagine a cascade being sustained in a mailing list?
The difference is: there's no single point of control (moderator, mailing
list root, nodelist) for unmoderated newsgroups. And no single point of
failure. Good or bad, this is a fundamental difference between the two
entities. This is independant of the user interface, the transport mechanism,
or even the server software. You can certainly set up a local newsgroup that
functions just like a mailing list with any of the news software packages.
But somewhere in the combination of the Usenet software, the Usenet flood fill
algorithm, and the net's worldwide distribution something quite different
from any of these has been created.
--
Peter da Silva `-_-'
Network Management Technology Incorporated 'U`
1601 Industrial Blvd. Sugar Land, TX 77478 USA
+1 713 274 5180 "Hast Du heute schon Deinen Wolf umarmt?"
James C Deikun
>It's a pity that the 'header-people' see everything in email. I believe
>this is why so many of them do not see the Usenet users' vastly
>different perspective. Usenet is very different from mailing lists.
>Gatewaying Usenet to a mailing list is like gatewaying an elephant to
>its legs.
>
>(Visualize the famous elephant picture, with different people feeling
>different parts of the beast. The header-people see only the legs.)
I am on a number of mailing lists, and read a number of newsgroups.
I read this forum of discussion as comp.mail.headers. I see no
fundamental difference between Usenet and mailing lists, only a few
details. This is not owing to lack of a Usenet user's "vastly
different perspective." Your ad hominem argument fails to hold any
more water than most such do.
--
James Deikun, University of Pittsburgh
#include <std_disclaimer.h>
Quick, someone mention Hitler!
John Stanley
>Just out of curiosity, do you not have groups of people with common
>interests where you work or do they have a better way than email to
>communicate?
Of course there are groups with common interests. Face to face is
generally how they communicate. Why should they have to stoop to email?
James C Deikun
John Stanley <sta...@skyking.OCE.ORST.EDU> wrote:
>In article <IiezJUSSM...@transarc.com>,
> <Craig_E...@transarc.COM> wrote:
>>A truly broken gateway.
>
>No, a properly working gateway. Gateways that claim that I mailed a
>message when I did not are broken. Gateways that don't tell you that
>mail came from them are hiding the fact that the mail came from them.
This sort of information was what the Sender: header was meant for.
The From: header and the Reply-To: header (if any) should still point
to you, since gatewayed or not the message is still FROM you.
All private reply functions on mailers I've seen use Reply-to: if
present, From: if not. With a proper gateway this should not break.
>That's right, but it happened. But when there becomes a "standard" that
>says that a "newsgroups" header means mail is supposed to be (or
>already was) posted, you will be hard pressed to say that that same
>transport is broken. "That header says it will be posted, so what's the
>problem if I do it here?"
There is no problem. Where else would the responsibility for posting
it lie?
>>What problem are we trying to solve? Coding around a broken transport
>>layer that posts rather than mails?
>
>No. Preventing things from breaking when someone creates a new meaning
>for existing headers.
We are, you aren't. You're trying to make sure no one creates a new
meaning for existing headers. You have offered absolutely no solutions
to the problems that might occur if someone does, yet you have made no
convincing arguments that these problems are insoluble. You can only
point to broken implementations. This shows absolutely nothing other
than your pessimism and your ability to whine. Make an argument with
generality and maybe you'll convince someone.
--
James Deikun, University of Pittsburgh
#include <std_disclaimer.h>
We tend to scoff at the beliefs of the ancients. But we can't scoff at
them personally, to their faces, and this is what annoys me. --Jack Handey
Leslie Mikesell
In my organization they tend to be located in different states, often
one person per state. We can only afford face-to-face a few times
a year.
Les Mikesell
l...@mcs.com
Mark Crispin
news are somehow different worlds. Throughout this entire exchange, I
have particpated solely as a member of header-people. Obviously, there is
now a newsgroup that gateways header-people mail, and Stanley does not
realize that when he posts to that newsgroup he also send mail to a wider
audience.
Nonetheless, I don't like getting flamed for replying to mailing list
mail, even if the flamer does so out of ignorance.
On Mon, 24 Oct 1994, John Stanley wrote:
> Foo yourself. I don't give a shit if you are impressed or not. I have
> asked you not to dump this stuff in my mailbox, and to let it wait until
> I have time to deal with it in a proper manner -- that is, when I get
> time to read news and have time to reply to your arguments.
Are you aware that you are ``dumping stuff'' into my mailbox? Has it
dawned on you that you are SENDING MAIL, to the header-people mailing list?
I am not reading this thread in news. I am reading it in mail, on the
header-people mailing list, and I send a mailing list response. Live
with it.
> You have continued to ignore that request. Why, I don't know. Maybe you
> don't understand simple english.
Once again, you claim that I ``don't understand simple english''. This
time, you lace your message with vulgarities. I have no interest in doing
anything to make life easier for you until you learn a more civilized form
of discourse.
> You have chosen to participate in mail. That is your right. I have chosen
> to participate in news. That is my right. Now, stop mailing me this crap.
> Is that clear enough for you?
I understood perfectly well what you are saying. Now, try to understand
why you are not going to get your wish:
I am not going to keep a list of special people who post to mailing lists
but are not to get replies, and manually remove such people each and every
time I send a reply. I will use my mailer's reply command. If you don't
want to receive replies to mailing list messages that you post, there are
headers that accomplish this function. Read RFC-822.
> You sent me the first email of this exchange.
Wrong. Get off your high horse. You have sent email to the header-people
mailing list for quite some time. Today, there were 5 or 6 messages from
you in my mailbox, all sent to the header-people mailing list.
You, sir, engage in the very activity that you so vehemently denounce.
> I made the mistake of
> replying to it. I tried to rectify that mistake, but you continue to
> ignore a simple request to remove my name from your mailings. Why?
Your demand is inane. It indicates that you can't be bothered to learn
what you are doing, or that not all environments match yours. The burden
is on you to know what you do, and to learn that not everyone uses the
same system as you.
Just maybe, if satori has dawned, you may have come to understand what is
happening, that you have been flaming me without just cause, and that you
have been a jack-ass. You may then receive the next bit of enlightenment,
which is that your model of news and mail is false.
You have demonstrated very effectively the incorrectness of your
understanding of news and mail. The situation which arose (and I know
quite well how and why) did so due to mechanisms completely unrelated to
the handling of the Newsgroup: header. Your proposed solutions would not
have prevented it in the slightest.
Yes, I am being hard-nosed at this point. But you have been a nasty
insulting jack-ass with no provocation. You would be best off realizing
that you have been a jack-ass, and apologizing for your insults.
The fact that this exchange occurred indicates that there is an
email/news integration problem of much greater depth than has been
examined in this thread so far. You perhaps, were a victim of this
problem; but so was I, and I resent you dumping your anger on me.
If you can admit your mistake, then you can start being part of the
solution instead of continuing to be part of the problem.
Keith Moore
> Broadcast entities, particularly distributed broadcast entities, are open
> in a way that even systems like fidonet echoes don't see... and don't have
> to deal with. Can you imagine someone posting
>
> I AM THE TERROR THAT FLAPS
> NO CARRIER
>
> multiple times to multiple fido echoes, or mailing lists, or moderated
> groups for very long before the offending site was cut out of the feed?
Actually, I can, at least on a mailing list. Especially on a peered list
that happened to strip Received headers. (there used to be a lot of them).
> Can you imagine a cascade being sustained in a mailing list?
Yes, on a large unmoderated list. I've actually seen this happen (though
probably not to the extent it happens on usenet).
> The difference is: there's no single point of control (moderator, mailing
> list root, nodelist) for unmoderated newsgroups. And no single point of
> failure.
There are other differences that contribute. The number of participants in a group
has a lot to do with it. It only takes one person to make a group useless, and
the larger a group, the greater the probability that there will be such a person in
the group. (For major usenet groups this probability approaches 1.) But this
happens on large mailing lists also.
Also, the lower the signal to noise ratio in a group, the less valuable
it is, so there's less effort expended to keep the group membership in line.
Not that I completely disagree with what you are saying, just that the locus
of control is not the only important factor.
> But somewhere in the combination of the Usenet software, the Usenet flood
> fill algorithm, and the net's worldwide distribution something quite different
> from any of these has been created.
...just as every mail system is different than every other one. A LAN mail system
that provides (e.g.) true receipt confirmation, cancellation of messages, reply to
sender only (no ability to reply to sender and all recipients), will produce a very
different set of expectations (== different "culture") than (e.g.) the Internet
mail system. Just because we call them both "email" doesn't mean they're the same.
No doubt there are some who would argue that we should not connect LAN mail systems
to the Internet, because of the culture clash that results. But there are many
incentives to do so, and eventually we will end up refining the Internet mail
protocols to accomodate LANs (e.g. new MIME content-types, delivery notifications,
etc.) and perhaps re-defining expectations of service for both (give up some LAN
mail features that don't work in a WAN environment, but add other features that are
feasible).
Keith
Mike Schenk
>I see absolutely no reason why I should have to care whether something is news
>or mail. It's up to these expensive computers to figure that detail out.
>
>Ideally, there would be no Newsgroups: line. Newsgroups would appear on the
>To:, cc:, or bcc: lines where they belong. But we live in an imperfect world.
>The actual integration effort is much more modest in its scope and much less
>disruptive.
No, I don't agree, the semantics of a Newsgroups: header is completely
different from a To: header (or Cc: or Bcc:).
If you write a letter (on paper I mean) you put To: Recipients on top of
it. That's because you're actually sending it to these people.
News is more like a bulletin board (like they have in supermarkets). If
you put up a note in the supermarket to sell your bike, you don't put
"To: everyone who wants to buy my bike" on top of it. You just stick it
in the right area of the bulletin board. And the Newsgroups: header is
basically an identification of the right area.
Now people are arguing, "how about mailing lists?". Basically, mailing
lists are used when it's not possible to create a Newsgroup (not enough
people interested or any other reason). So a mailing list is basically a
"hack" to allow a news-like exchange of messages when for some reason, a
newsgroup is not possible. This does not mean that email in general is
the same as news.
Once again, I have no problem with people who use the same frontend
application for mail and news but I have a severe problem if people
start changing the meaning of mail and news headers just because they
don't see the difference that are people see. This would mess up things
a lot for people who see the differences.
Mike
Ned Freed
> that provides (e.g.) true receipt confirmation, cancellation of messages, reply to
> sender only (no ability to reply to sender and all recipients), will produce a very
> different set of expectations (== different "culture") than (e.g.) the Internet
> mail system. Just because we call them both "email" doesn't mean they're the same.
There are also quite a few mail systems that maintain a single copy of a given
message and provide multiple recipients with pointers to the one copy. However,
Internet use tends to associate use of this model more with news than with
email. Users who are accustomed to this model for email routinely send messages
to thousands or even tens of thousands of recipients, and have great difficulty
when faced with email systems that make separate copies and break down under
this sort of load.
There are also systems other than news that include almost all of the
capabilities of news as a subset. Lotus Notes is one example -- there are
others.
> No doubt there are some who would argue that we should not connect LAN mail systems
> to the Internet, because of the culture clash that results. But there are many
> incentives to do so, and eventually we will end up refining the Internet mail
> protocols to accomodate LANs (e.g. new MIME content-types, delivery notifications,
> etc.) and perhaps re-defining expectations of service for both (give up some LAN
> mail features that don't work in a WAN environment, but add other features that are
> feasible).
Very true. However, it is interesting to note that it is far more common to
encounter problems with limitations on the LAN email side than it is with
limitations on the Internet side. Classic examples of this include lack of
support for the full range of originator information in messages (i.e. only one
or two fields to handle all of from, resent-from, sender, resent-sender,
reply-to, resent-reply-to, and envelope from), lack of proper separation
between envelope and header, and lack of support for structures more complex
than one text object and a set of unlabelled binary attachments.
Ned
Rahul Dhesi
<jg...@CMU.EDU> writes:
>The address "arpalists+h...@andrew.cmu.edu" is subscribed to
>the header-people mailing list. Any mail sent to this address gets
>delivered to the folder named "internet.header-people", where it can
>be read by anyone on the system.
Ok, so far arpalists+h...@andrew.cmu.edu sounds like a
reimplementation of Usenet.
What happens if I cross-post something to both comp.mail.headers and
comp.mail.sendmail, because it's something to do with headers as parsed
by sendmail, and I set followups to comp.mail.sendmail, as I would like
the discussion to limit itself to sendmail's specific behavior?
Now does your reimplementation of Usenet still work now? If it does, I
would like to know how this is achieved.
And what if one of the followups to my posting happens to be a flame,
and out of courtesy, the poster cross-posts it to alt.flame and directs
followups there?
What if you send email to some list server somewhere asking it to
subscribe jg...@CMU.EDU to it? The list server might send its output
with the To: line showing some list...@somewhere.edu. How does the
software at CMU know to put it in a shared message base? If it
doesn't, have you really achieved any News/mail unification?
--
Rahul Dhesi <dh...@rahul.net>
Keith Moore
That's a funny assumption. I have about the same number of incoming
mail folders (addressed to moore+fo...@cs.utk.edu) as I do
newsgroups that I'm subscribed to. But when I have a choice about
how to get access to a particular discussion, email always wins over
news. The reason? email doesn't expire, so it's very easy for
me to keep logs of important discussions.
On the other hand, I am repeatedly deluged with requests on one
of the lists that I run, to make it into a newsgroup. The reason?
most people don't get email pre-sorted into folders, but they do
get news pre-sorted into newsgroups which they can read at their
leisure.
What this tells me is that the main differences between mail
and news, as far as users are concerned, is in the user interface --
not any intrinsic property of either service.
That's not to say that a newsgroup is exactly the same as a mailing list.
It's not. But I don't believe that users care about the finer points
of either. Given a choice, they will pick the one that provides the
best overall interface, *including* the user agent.
--
Keith Moore NETLIB development group
Computer Science Department / University of Tennessee at Knoxville
107 Ayres Hall / Knoxville TN 37996-1301
Let's stamp out software patents in our lifetime.
Ned Freed
> this is why so many of them do not see the Usenet users' vastly
> different perspective. Usenet is very different from mailing lists.
> Gatewaying Usenet to a mailing list is like gatewaying an elephant to
> its legs.
Reality check time. I know quite a few of the "header-people" involved in this
particular discussion personally, and I know that they have extensive
experience with both news and email. I know I do -- I use both on a regular
basis and have done so for many years.
I happen to read header-people as a mailing list because the discussion is very
important to the work I do and I want to be able to construct my own archives
and not worry about article expiration. As a matter of fact I read this
discussion using news for several years, but various factors conspired to force
a change to email.
I also read many other discussions using a news interface -- always have and
probably always will. And it happens to be a different interface from the one I
use for email. (I'd like for it to be integrated and failing that I'd love to
have the MIME support I have for email in my news interface, but neither has
happended yet.)
If there's a vastly different perspective here, as a longtime news/email user I
do not see it. There are large differences between news and email, but most
users only pay attention to them when they are forced to by implementation
artefacts.
Ned
Ned Freed
> 1) Reply personally to sender only
> 2) Reply to sending medium (the list, or newsgroup) only
> [ ignore cases of mixed medium messages for now ]
I disagree. When I'm engaged in a serious discussion I want my replies to go
directly to the people directly involved in the discussion. I also want a copy
of my reply to go back to the list.
There are at least two good reasons for this:
(1) Time. Modern Internet email frequently lets me carry on world-wide
discussions in near real time. When things are really cruising I often
see messages sent every minute. But I can only get these sorts of response
times from direct personal mail. Lists take time to expand -- big lists
take longer and news longer still -- time that I frequently can't afford
if deadlines are to be met. Nevertheless, I also want the messages to go
to the list so that they can be reviewed by the large lurking audience
out there.
(2) Insurance. How many times have you seen someone say, "Please reply to
me directly since I'm not a member of this list." Or "Please reply to
me directly since I don't read this group regularly." The simple
existance of such requests means that agents have to be able to reply
to both the sender and the medium at once. But I also don't trust
my own ability to know when such a request is in effect -- sometimes I
miss the request and sometimes the sender didn't bother to mention it.
I therefore reply to both the sender and the list as a matter of course.
> Unfortunately, I'm not aware of any MUA that does (2) correctly.
Well, I'm not exactly sure what you expect an MUA to do in this case, but I'd
say that quite a few of them do a pretty good job as long as only one medium is
involved. In the single medium case the real problem is that there is a large
amount of inconsistency in how message headers are constructed on various
lists. In quite a few cases the list address is nowhere to be seen in the
headers, which makes it awfully hard for an MUA to reply to an address it
doesn't know.
As for the problem of multiple mediums, isn't this the issue we're discussing
in a nutshell?
> John Stanley's anger and flamage is a result of the broken system
> we have that can't do the right thing. And he does have a point.
> He (rightly) assumes that discussions take place in the medium
> he's posting to, even though other people are getting things by
> other mediums. The system should treat things transparently, but
> it doesn't.
I agree that it would be nice for this to be handled transparently. But this
doesn't automatically lead to the outcome.
Let's suppose that the gateway for this list to and from news was seamless and
transparent, and that all the MUAs and NUAs always did the right thing. Even if
this were true there would still be a "problem", since I for one would be
responding to both the sender and the medium more or less as a personal
default. I would even do this if there was only one medium involved, because
this is the way of working that I've found produces the best results overall.
> What MUA's do for (2) is reply BOTH to the sender AND to the sending
> medium.
I can't speak for other MUAs, but this certainly isn't true for the one I use.
It defaults to sending the reply to the medium only in most cases (modulo the
availability of information about the medium, of course) and to the sender in
others. (It depends on what the headers say.) It never defaults to sending to
both. (Well, there's this system in Sweden that builds up a list of every
individual, list and newgroup involved in the discussion, creating aliases
on-the-fly as necessary, and puts them all in *all* of the various headers,
thus insuring that everyone gets everything no matter what, but it is a very
special case.) I have to specifically tell my MUA to send to both, and I've
already outlined my reasons for doing this almost all of the time.
> Bah. Now the person gets TWO copies via different mediums
> and he (rightly) assumes if it was e-mail that it was a personal
> reply, and if it was news that it was an open reply.
Right. But your assumption that I don't intend for there to be a private copy
is incorrect. I find this to be very useful.
> So you say "oh, just indicate to the system how to direct replies
> appropriately". Too bad you can't.
I agree that this doesn't work well for mixed mediums.
> Fudging Reply-To is wrong because that requires that you know the
> correct e-mail gateway if you're posting to news, and it also assumes
> that the gateway is bi-directional. (and it assumes the Reply-To e-mail
> address will work for everyone, or the system will properly pre-munge
> it before giving it to the recipient)
> Even assuming Reply-To does work, there's no standard "Reply" method,
> and some MUA's will still insist on Cc'ing the original sender
> personally. (arguably they are broken, but they do exist)
Equally arguably they are working correctly and the other ones are broken.
> I try whenever replying to mail from a list, to edit the headers to not
> Cc the sender, unless I'm not sure the person is on the list or needs
> an urgent reply.
Uh huh. Exactly my point. I'm *never* sure that the person is on the list or
will receive a reply directed to the list in a timely fashion. That's why
I almost always insist on sending a separate copy directly.
Ned
Craig_E...@transarc.com
e-m.. John Sta...@skyking.oce (1654)
> >My advice would be to abandon the mail/news distinction, and instead
> >push MUA developers to ensuring that users know the difference between
> >reply-to-sender and reply-to-all/reply-to-recipients.
> And if the "sender" is a mailing list or USENET newsgroup? People who
> think "reply to sender" is a private message and thus dump their life
> histories into it, only to find that "sender" was a mail-news gateway
> are going to be rather unhappy. "Hey, that was supposed to be private! I
> used the private reply funtion!" "No, you used 'reply-to-sender'. There
> isn't any 'private reply' function."
A truly broken gateway. By the way, ``sender'' wasn't meant to mean
contents of the Sender: field. It was meant as the From: address (well,
Reply-To: if that's present). It's unfortunate that this is garbled by
gateways, too. (Clearly, a message sender needs to be reminded of where
a message is to be sent, before giving the final go-ahead.)
> >on the To: field. Given that the prefix is ``netnews.'', it's unlikely
> >to be part of a legitimate user's name. Should that situation arise,
> >the MUA will query the message-composer as to which recipient was
> >intended.
> The recipient may very well have no choice. If a "helpful" mail
> transport decides that it should post the message instead of passing it
> to the system in the To: address, because it is more efficient, what
> matters if the sender intended it to be a private message?
It's the sender who has the choice. I don't understand this at all.
If the mail transport is posting rather than mailing, it's broken.
> >> It is not acceptable to me.
> >
> >What would you propose, that doesn't presume that all mail is private?
> Create a new set of headers, that are NOT already defined to mean
> something else. For example: Also-mailed-to: and Also-posted-to:.
> Of course, I suggested these already.
What problem are we trying to solve? Coding around a broken transport
layer that posts rather than mails? Isn't this the wrong level at which
to attack the problem?
Craig
James C Deikun
>The correct choice, when discussing email vs News, is between 'send
>private reply to author' and 'post public follow-up'. A follow-up is
>never sent to any specific people. A follow-up, from the point of view
>of the person posting it, has no specific recipients. Anybody who
>chooses to read that follow-up -- and there is no predefined list of
>such people -- becomes a recipient. Usenet is far more dynamic than
>may appear at first sight.
>
>You might call this a trichotomy:
>
>1. Send private response to original author.
>2. Send public response to all recipients, if you can identify them.
>3. Send public follow-up to various news spool directories, where
> countless unidentifiable people will have access to it.
This is an obvious straw man argument. Very few people who send to
subscribable mailing lists have the slightest knowledge of the
identities of the lurkers who may receive the mail. Some lists are
exclusive enough that you have some real idea who your readers will
be, but some will let anyone subscribe who has a valid mail address.
Some are gated to news or to BBS fora, where they end up with the
same kind of audience as normally posted news. Mail already embodies
a full spectrum ranging from public to private. Newsgroups, OTOH,
are mostly clustered around the public end. Some newsgroups,
however, never leave a site. Newsreaders already also have some
degree of integration with mail, which is how you can even make a
private reply. The interfaces for news and mail would be somewhat
changed by the integration; IMHO to the better. Their fundamental
natures, though, would not be compromised at all since they already
overlap.
John Gardiner Myers
> In <QifGPSy00...@andrew.cmu.edu> John Gardiner Myers
> <jg...@CMU.EDU> writes:
>
> >The address "arpalists+h...@andrew.cmu.edu" is subscribed to
> >the header-people mailing list. Any mail sent to this address gets
> >delivered to the folder named "internet.header-people", where it can
> >be read by anyone on the system.
>
> Ok, so far arpalists+h...@andrew.cmu.edu sounds like a
> reimplementation of Usenet.
Once inserted into the folder "internet.header-people", it does no
flooding to other systems.
Using the same mechanism, any mail sent to the address
"jm...@andrew.cmu.edu" gets delivered to a different folder, where it
can be read by anyone on the system that jm36 gives access to. Does
this sound like a reimplementation of Usenet?
Similarly, jm36 could arrange to have any mail sent to the address
"jm36...@andrew.cmu.edu" get delivered into some other folder and
give read and/or write access to that folder to any arbitrary subset
of the users on the system.
> What happens if I cross-post something to both comp.mail.headers and
> comp.mail.sendmail, because it's something to do with headers as parsed
> by sendmail, and I set followups to comp.mail.sendmail, as I would like
> the discussion to limit itself to sendmail's specific behavior?
> Now does your reimplementation of Usenet still work now? If it does, I
> would like to know how this is achieved.
On the netnews side, the part that copies the messages into the folders
netnews.comp.mail.headers and netnews.comp.mail.sendmail sees the
Followup-to: header and generates an internal header
"X-Andrew-Widereply:" directing the "Reply to Readers" function to
send only to netnews.comp.mail.sendmail.
On the arpalists+ side, the part that files the message into
internet.header-people doesn't know to look for the Followup-to:
header.
This is a less than desirable situation for several reasons. It would
be better for the clients to look for the Followup-To: header themselves.
> What if you send email to some list server somewhere asking it to
> subscribe jg...@CMU.EDU to it? The list server might send its output
> with the To: line showing some list...@somewhere.edu. How does the
> software at CMU know to put it in a shared message base? If it
> doesn't, have you really achieved any News/mail unification?
What does the address in the To: line have to do with delivery? It's
the *envelope* recipient addresses that matters.
Mail to the envelope address jg...@CMU.EDU gets forwarded to the
envelope address jm...@andrew.cmu.edu, which in turn gets filed into
my "mail" folder.
I wouldn't ever send mail to a list server asking it to subscribe
jg...@CMU.EDU--I never subscribe my personal mailbox to mailing lists.
Depending on the list, I would either send mail to the appropriate
administrator asking them to create a new internet.* folder and
subscribe it to the list, or I would set up a restricted-access folder
and subscribe it to the list. (In one extreme case, the
restricted-access folder is restricted to just me.)
John Stanley
Mark Crispin <m...@Panda.COM> wrote:
>Perhaps this is a good example to debunk the foolish notion that email and
>news are somehow different worlds.
Mark, will you please stop and listen to other people? To you they are
the same because you want them to be. That does not make them the same
to everyone else. I think it should be very clear to you by now that
they are NOT the same to me, and are NOT used for the same purpose.
>Throughout this entire exchange, I
>have particpated solely as a member of header-people.
By your own choice, I assume.
>now a newsgroup that gateways header-people mail, and Stanley does not
>realize that when he posts to that newsgroup he also send mail to a wider
>audience.
No, I know very well that the group is gatewayed.
>Nonetheless, I don't like getting flamed for replying to mailing list
>mail, even if the flamer does so out of ignorance.
You are getting flamed for replying in a way you have been asked not to.
Reply away, except stop sending it to me. I am not on your mailing
list. Don't force me to be.
>Are you aware that you are ``dumping stuff'' into my mailbox? Has it
>dawned on you that you are SENDING MAIL, to the header-people mailing list?
You have chosen to receive this mail. If you wish not to receive it,
then you need to remove yourself from the mailing list. Whatever you
receive by mail you are receiving because you want it there. If you
were to decide tomorrow that you did not want to work in mail, but
instead participated by news, you could make that choice, and you would
receive NOTHING further from me by mail. Yet, you refuse someone else
the simple courtesy of being able to make the same decision you expect
for yourself.
>I am not reading this thread in news. I am reading it in mail, on the
>header-people mailing list, and I send a mailing list response. Live
>with it.
You are not sending a mailing list response. You are sending a response
TO ME, with a CC of the mailing list. I have asked you multiple times to
stop doing that. Reply to the list if you want, but stop replying to
me.
>Once again, you claim that I ``don't understand simple english''. This
Yes, it's now more than clear that you are doing this because you are an
asshole. Oh, did that language offend your sensitive eyes? Tough. Honor
the request I made and I will stop calling you an asshole.
>time, you lace your message with vulgarities. I have no interest in doing
>anything to make life easier for you until you learn a more civilized form
>of discourse.
I have no interest in being civil to you until you respect me enough to
honor the request I made. In case you were wondering, it doesn't take
much.
>I am not going to keep a list of special people who post to mailing lists
>but are not to get replies, and manually remove such people each and every
>time I send a reply.
You really are slow, aren't you Mark? You don't need a special list.
Just understand that when you send mail to "header-people", it is
gatewayed right back into news where it will reach me just fine. You do
rememeber the gateway you were lecturing me about? Well, it goes both
ways. And, I would dare say, anyone else who posts to news to whom you
reply will gat a copy just like I would.
And, I would guess that, were you to receieve mail sent directly to the
mailing list from another member, it would get back to them if you
replied only to the list. That how mailing lists work.
So, reply to the list. Your words will get out. Replies will go to
everyone who participates, even if they are doing so via news. One simple
reply. One choice for all replies to header-people.
>I will use my mailer's reply command. If you don't
>want to receive replies to mailing list messages that you post, there are
>headers that accomplish this function. Read RFC-822.
I should not have to insert special headers to get around a broken
gateway. (Yes, your own gateway should be inserting a "reply-to:
header-people..." That is where replies should be going.) I should not
have to know the submission address of the mailing list in order to keep
your crap out of my mailbox. That's one of those "integration" issues
you keep harping on. Why is the header-people mailing list not properly
integrating?
>> You sent me the first email of this exchange.
>
>Wrong. Get off your high horse.
Pull your head out ... I sent no mail until I replied to yours. Until
that time, I had posted to news. The gateway sent you the mail.
>You, sir, engage in the very activity that you so vehemently denounce.
That is incorrect. I am sending no mail to people who have not decided
for themselves that they wish to participate by mail. Every person on
the mailing list made their own decision.
>Just maybe, if satori has dawned, you may have come to understand what is
>happening, that you have been flaming me without just cause, and that you
You have given me cause every time you mail your responses to me instead
of to the mailing list.
>You have demonstrated very effectively the incorrectness of your
>understanding of news and mail.
I have demonstrated beyond any reasonable doubt that you do not have the
only viewpoint on the use and propriety of mail and news. I wish that
you would open your eyes to that fact, and accept that other people can
have a different viewpoint than yours.
>The situation which arose (and I know
>quite well how and why) did so due to mechanisms completely unrelated to
>the handling of the Newsgroup: header. Your proposed solutions would not
>have prevented it in the slightest.
You are so smart to recognize that this is now a different issue. It has
nothing to do with whether my arguments are right or wrong, it is a
simple matter of respecting a simple request.
>Yes, I am being hard-nosed at this point. But you have been a nasty
>insulting jack-ass with no provocation.
You have continued to provoke this issue with every mail you send. I
tried speaking to you harshly with the misguided hope that it would
wake you up and make you realize that I feel strongly about this.
Instead, you dug your heels in. Well, fine.
>You would be best off realizing
>that you have been a jack-ass, and apologizing for your insults.
No, I am best off realizing that you are not going to behave in a
civilized manner by respecting the wishes of others, and just writing you
off as a lost cause.
>The fact that this exchange occurred indicates that there is an
>email/news integration problem of much greater depth than has been
>examined in this thread so far.
It is not an integration problem. It is a problem in two places: 1) a
gateway that does not insert a "reply-to" pointing back to the gateway,
thus leaving YOU without the ability to select whether to reply To: me
or not, and 2) a jerk who won't accept anyone else's viewpoint even as
much as to honor a request to stop sending mail.
>You perhaps, were a victim of this problem; but so was I,
You were a victim of the gateway, the latter problem is your own making.
Had you honored the request I made, none of the flaming you whine about
would have happened.
>and I resent you dumping your anger on me.
I resent you sending this stuff to me in mail. See how far that has
gotten me?
>If you can admit your mistake,
The only mistake I made was replying to your mail the first time. I have
already admitted that.
>then you can start being part of the
>solution instead of continuing to be part of the problem.
The solution is for you to stop sendimg me mail. I can do nothing about
that. Simple english requests to stop have fallen on deaf or ignorant
ears.
Now, if you want to continue discussing this side issue, feel free. I am
writing you off as an "authority" until such time as you broaden your
world view to include other people's usage of mail and news.
Craig_E...@transarc.com
> Mailers that leave Newsgroups: headers in mail in reply to news articles
> are broken. Period.
OK, maybe I'm just confused, but I don't know why you think they're
broken. What transport would ever look at such a header without being
shot at sight? (And the RFC's don't back up this ``broken'' assessment.)
Oh, it's the NMUA (combined News/Mail User Agent) on the receiving end
that can't figure out how to do a wide-reply (public reply, e.g. reply
to recipients, or reply to everybody listed), since it will post as well
as mail? Presumably the NMUA's user has chosen to do one of these broad
replies, so there's no particular privacy being violated: whatever the
user chose to wide-reply, will be sent widely, but maybe more broadly
than was the message to which the reply was generated. Of course,
there's no reason why the NMUA has to look for the Newsgroups: header to
figure this out, when it could better get information out-of-band that
the message to which it's replying was received via mail, not via
netnews.
It's too bad that, in mail from a NMUA, there's nothing to say whether
that message was delivered only via mail, or via mail and netnews as
well; the presence of a Newsgroups: header obviously can't be counted
on. But only a wide-reply should ever pick up the Newsgroups: contents
in any event.
Users who intend to reply privately and use the reply-to-all (public
reply) kinds of commands to MUAs deserve what they get.
Craig
Craig_E...@transarc.com
e-m.. Keith Mo...@cs.utk.edu (1391)
> > It's a pity that the 'header-people' see everything in email.
> That's a funny assumption. I have about the same number of incoming
> mail folders (addressed to moore+fo...@cs.utk.edu) as I do
> newsgroups that I'm subscribed to.
In the AMS installation at andrew.cmu.edu, there are several public
users that represent the bboard system, and they get all the bboard mail
from all kinds of lists. (Header-People, for instance, comes in as
ARPALISTS#HEADER...@andrew.cmu.edu, but could just as easily arrive
for arpalists+h...@andrew.cmu.edu; it's filed into a public
folder in AFS. At Transarc, I can read the bboard from the
andrew.cmu.edu AFS cell. The management for this bboard is a matter of
andrew.cmu.edu administrative policy. It looks a *whole* lot like
netnews, but it's entirely done with mail. A comparable collection of
local bboard folders exists in the transarc.com cell as well.
Certainly I'd do the same thing with private subscriptions (e.g.
subscribe to them as cfe+SU...@transarc.com), but this collection of
local bboard folders affords far better management for most purposes.
Looks like a newsgroup, works like mail.
Craig
John Gardiner Myers
> Excerpts from internet.header-people: 23-Oct-94 Re: Harmonzing Internet
> e-m.. James C Dei...@pitt.edu (4360)
>
> > Mailers that leave Newsgroups: headers in mail in reply to news articles
> > are broken. Period.
>
> OK, maybe I'm just confused, but I don't know why you think they're
> broken.
The mailers are broken because they are using the same header name for
two completely different and incompatible semantic meanings.
Rahul Dhesi
Deikun) writes:
>Your ad hominem argument fails to hold any
>more water than most such do.
Not ad hominem at all. If you recall, Mark Crispin said:
>You have posted several messages to HEADER-PEOPLE, an email mailing list which
>has been around for over 17 years. Not only that, but after sending the above
>message you proceeded to post two additional messages to HEADER-PEOPLE, 22 and
>33 minutes later, as attached below.
Hmm...In fact the person he was addressing had been posting to Usenet,
but Mark assumed he had been sending email to the header-people mailing
list. Why? Possibly because, as I said:
It's a pity that the 'header-people' see everything in email. I believe
this is why so many of them do not see the Usenet users' vastly
different perspective.
I'm sending this posting to comp.mail.headers, but no doubt I will soon
be accused of sending email to header-people. Some of them will no
doubt send me duplicate responses, one copy as email cc'd to
header-people, one copy as a Usenet posting. And they will see nothing
undesirable about this sort of unnecessary duplication, which Usenet
avoids but mailing lists do not, and they will continue to insist that
Usenet is just a form of email.
...And should I crosspost this article to another newsgroup, and some
followups get moved into a different set of newsgroups, they will miss
this happening completely.
--
Rahul Dhesi <dh...@rahul.net>
Mark Crispin
> Hmm...In fact the person he was addressing had been posting to Usenet,
> but Mark assumed he had been sending email to the header-people mailing
> list. Why? Possibly because, as I said:
> It's a pity that the 'header-people' see everything in email. I believe
> this is why so many of them do not see the Usenet users' vastly
> different perspective.
Hi Rahul. I knew what was going on. I was flamed by an individual who
lept to a completely unwarranted conclusion about my environment. I
responded with how his behavior appears at my end. Even though I knew
how/what/why, one can not make the presumption that some other victim
would know why he is receiving such abuse from this individual.
It is not acceptable to pin blame on mailing lists. News is all fine and
good, but it is a luxury that not everyone shares. Email-connected users
greatly outnumber news-connected users. Users with email connections
only have no choice if they are to participate in online discussions.
It is *wrong* to flame someone for being at the other end of a dual
environment discussion. It is *wrong* to flame someone because the
deliver of his messages reflect the limitations of the medium. The key to
the matter is to recognize this, and not to make vulgar accusations.
> I'm sending this posting to comp.mail.headers, but no doubt I will soon
> be accused of sending email to header-people.
I doubt that anyone would accuse you of sending email to header-people,
because you don't send vulgar messages accusing people of stuffing your
Very Important email inbox!
-- Mark --
Steven D. Majewski
I often use email to communicate with people just down the hall in my
department: email discussions leave a "paper(less) trail" that is often
useful. And even with people in the same office - with different
schedules and meetings to match, you can often end up playing "intercom
tag".
Of course face to face is better bandwidth, and there are times it can't
be beat - but often after a F2F meeting, we exchange email notes about
what we each believe we agreed to!
Of course we could exchange paper memos or meeting minutes, but
paper memo's have to be walked down to the mailbox and can have >24 hour
lag - depending on how frequently people check their mailboxes. Paper
memo's have to be filed and can't be "grep"-ed for.
So I wouldn't call it "stooping" to email!
-- Steve Majewski (804-982-0831) <sd...@Virginia.EDU> --
-- UVA Department of Molecular Physiology and Biological Physics --
-- Box 449 Health Science Center Charlottesville,VA 22908 --
Ned Freed
You must operate in the most tyrannical of working environments if you actually
see things this way! Everyone must work the same hours exactly, attend all the
same meetings, use the phone at the same time, never get sick, never have any
children (they get sick -- made me two hours late today), go to the bathroom
in tandem, and nobody is allowed to work or even think about anything
job-related after hours.
I work for a company where different employess work very different hours, both
as a matter of personal preference (I come in late and leave late, others come
in early and leave early) and as a matter of necessity (we have people working
in at least seven different time zones). We also send people to a large number
of trade shows, so it is rare for there to be a time when everyone who works at
this office to be here at the same time.
As a company we often deal with extremely complex software development issues,
issues where it is essential that people sit down and think things through, and
write it all down before a meaningful discussion can take place. We also spend
considerable time doing technical support for our customers, and being able to
track problems and problem fixes and enhancements via group communication
mechanisms is a crucial part of getting the job done.
If I didn't have email and news and the private group facilities they provide I
could not be 1/10 as productive as I am. I frequently send email to someone
sitting five feet away because they are on the phone and I don't want to
interrupt them.
I like to think that even if we didn't develop and sell products in this area
we'd be just as intelligent in our use of them. Its probably just a fantasy on
my part, but then again I've encountered more than one company whose focus
has nothing to do with computer hardware and software who uses these sorts of
tools even more extensively than we do.
Ned
Roger Fajman
>
> 1) Reply personally to sender only
> 2) Reply to sending medium (the list, or newsgroup) only
I disagree. In my opinion, when there's more than one possible
place to reply to, the user should be forced to choose which to
use. There should not be a default and all combinations should be
allowed.
Rahul Dhesi
Moore) writes:
>> It's a pity that the 'header-people' see everything in email.
>That's a funny assumption.
Here is my email reply which I sent to Keith.
-- Rahul
Date: Wed, 26 Oct 94 03:08:41 PDT
From: Rahul Dhesi <dhesi>
To: Keith Moore <mo...@cs.utk.edu>
Subject: Re: Harmonzing Internet e-mail and Usenet News protocols
More a conclusion than an assumption. For one thing, just about all
the responses I see in email are cc'd to header-people, as was yours.
For another, people continue to talk about 'reply all' vs 'reply
sender' thus showing a heavy bias towards the mailing list vs private
email dichotomy. The 'reply all' terminlogy is nonexistent and
meaningless in a Usenet context.
Rahul
> Date: Wed, 26 Oct 94 00:42:46 EDT
> From: Keith Moore <mo...@cs.utk.edu>
> To: Rahul Dhesi <dh...@rahul.net>
> Cc: header...@mc.lcs.mit.edu, mo...@cs.utk.edu
> Message-Id: <1994102604...@wilma.cs.utk.edu>
> Subject: Re: Harmonzing Internet e-mail and Usenet News protocols
Craig_E...@transarc.com
e-m.. John Sta...@bubbles.wes (3969*)
> With even the current glaring dichotomy between public and private, there
> are still too many "how do I join this group" etc. messages. It can only
> get worse when the division becomes less obvious.
I'm glad to see that folks have adopted the public/private dichotomy,
but it's disheartening to see the news/mail dichotomy discussed in a
connected way in the same message. Where I live, most mail is in fact
public--shared among people in named groups.
My advice would be to abandon the mail/news distinction, and instead
push MUA developers to ensuring that users know the difference between
reply-to-sender and reply-to-all/reply-to-recipients. And, as possible,
the consequences of replies directed at unbounded sets of recipients.
FYI, the (implemented) model I've had uses a prefix for a newsgroup name
on the To: field. Given that the prefix is ``netnews.'', it's unlikely
to be part of a legitimate user's name. Should that situation arise,
the MUA will query the message-composer as to which recipient was
intended. The mail interchange formats don't have this problem; the
addresses are transformed to something unique, that cannot be part of a
user's name.
> But you have already said that you don't care about the
> difference between public and private, and, in fact, I don't believe that
> you have even gone as far as to admit that there is a difference. Thus,
> it is acceptable to you for this difference to be hidden.
> It is not acceptable to me.
What would you propose, that doesn't presume that all mail is private?
Craig
John Gardiner Myers
> News is always a public medium.
This is hardly the case. There are news groups restricted to certain
sites or sets of sites and some news implementation can restrict
groups down to individual users.
John Stanley
>In article <386apd$9...@gaia.ucs.orst.edu>,
>John Stanley <sta...@skyking.OCE.ORST.EDU> wrote:
>>>user's response level. It is appropriate to respond back to the group
>>>that originally saw a message and quote the original in the body. It
>
>>It may or may not be. For example, someone mentioned sending "how do I
>It's inappropriate to send such a message to a group in the first place.
Sigh.
Of course. You know that, I know that. I already replied to that effect.
But, it happens. And that is just one example of a message that is
NOT appropriate to reply back to the group that originally saw the
message.
>>So? They type of reply, and what it says, will often depend on the
>>nature of the message being replied to. To blur the difference between
>>public and private messages is to open the gate for people to think they
>>are replying privately when posting, or publicly when mailing privately.
>
>Sorry, but I don't blindly assume that everything in email is private,
>so there is nothing blurring any distinction here.
I didn't say everything in mail is private. I am talking about the
distinction between public and private. News is always public. Mail is
a mix. But everyone here seems to be dealing with mail only as public.
It isn't.
>>>is a judgement call as to whether or not it is appropriate to add new
>>>recipients or groups to the response when the original had more than
>>>one recipient.
>>
>>This is not the issue.
>
>It is exactly the issue. Nothing other than my own judgement can or
>should determine the type of reply I send.
You must base that judgement on facts at hand. If the fact that you are
replying in a public posting is lost, you may make the wrong decision.
Similarly, if the fact that you are replying in private is lost, then
you may decide incorrectly again.
Adding recipients is not the issue. You are free to add recipients to
private messages you send as you wish. That is why that is not the
issue.
>>And making it all look the same removes the perceived limitation that
>>you say is never appropriate.
>
>On the contrary, there is no limitation, only a choice.
Something that is never appropriate is a limitation. It is not a
physical one, but it is an ethical one.
>Making it look
>the same emphasizes the fact that you always have the same choice and
>you should make it appropriately.
Making it look the same removes information you need to know in order to
choose. It may emphasize that you need to make the choice, but it takes
away your ability to do so.
>Ummm, the user should *direct* the choice of replying to only the sender
>or to all the recipients. If they can't remember which they chose, it
>probably won't help to print the list again.
Right. Seeing a list of 80 names in a CC: header won't remind anyone
that they are replying in an essentially public manner. Seeing that they
are sending a response to "foobar-list" won't remind them that they
selected a public response. You must have some really dense users.
>It's not really rude, since it does call your attention to the response
>which might otherwise get lost since no one can possibly read all
>the newsgroups every day.
It wastes your time and dumps a public discussion into a private
mailbox. The reason news exists is to hold discussions in public. The
reason people post there is to discuss in public.
>It would just be nice if there were a
>standard way of indicating in the headers that the message was also
>posted to the newsgroup.
As long as that header is NOT a standard news header, I have absolutely
NO PROBLEM with this. I have even said as much, in public. I have even
gone as far as suggesting names for the new mail headers.
The problem I have is when people want news headers to play double duty.
There is no guarantee that some helpful mailer will not twist the
presence of a "Newsgroups" header in mail into "I must post this mail
for the user." In fact, I can almost guarantee that it WILL happen.
I have found at least one "helpful" mailer that made just such a
decision based on the USER NAME the mail was addressed to, completely
ignoring the tiny detail that the mail was addressed to a machine in a
completely different domain. If you start using a header that some
mailers leave in the mail they send in reply to news articles, you are
guaranteeing that some helpful soul will program his mail system to obey
it when it was not the intention of the user for it to be obeyed, nor
his choice to include it.
>But, if you are responding to a list message with
>an email interface, your own mailer will process the CC: address for
>your reply, although it may be stripped by the mailing list processing
>and prevent later responses from getting back to the added address.
And if one of those added addresses happens to be the person who asked
the question and who is expecting the replies to come to him instead of
him going to where the replies will be ...
Brian Edmonds
>>>>> "JS" == John Stanley <sta...@skyking.OCE.ORST.EDU> writes:
JS> I didn't say everything in mail is private. I am talking about the
JS> distinction between public and private. News is always public. Mail
JS> is a mix. But everyone here seems to be dealing with mail only as
JS> public. It isn't.
So if mail is a mix of public and private communication, how does
integrating news into it change that? It seems to me your campaign time
would be better spent encouraging people toward a better UI design that
ensures the user is aware of the public/private distinction rather than
trying to convince us that the current news/mail implementation embodies
this distinction.
---Ah!-Megami-sama---Greenwood--KOR--DP--Riding-Bean---/================
Brian Edmonds (Prodigy Technologies Corporation) / SCA: Bardas,OFLT
http://www.cs.ubc.ca/spider/edmonds ...treasury / Lions Gate, AnTir
---Talking-Heads---Cranberries--Beethoven--ZZ-Top---/___inux spoken here
John Stanley
On Thu, 20 Oct 1994, Mark Crispin wrote:
> John Stanley's argument basically boils down to:
I would appreciate it very much if you did not attempt to "boil down" my
arguments for me, especially if you are going to overlook major parts.
> The perceived models of Internet email, as established in common email
> software, and of USENET news, as established in common news software, are
> fundamentally different.
No. That is not what my argument is. The "established in" parts are
irrelevant. The perceived models are based on the difference between
public and private. Email is generally a private communication medium.
"Take it to email" is the cry in public groups, when a topic wears thin.
(Email is a mix of public and private, but this is not an excuse to
ignore the private side of the medium.)
News is always a public medium.
It is the same difference as between US Mail and the bulletin board down
at the grocery store.
> There is no way for these two models to integrate
> without creating an integrated model that has differences from the two. There
> are different protocols to transport mail and news.
Irrelevant.
> So far, he's correct.
Yes, but not because I am making the straw argument you made up for me.
> But, from this point, he makes this conclusion:
>
> This state of affairs is right and desirable.
Oh, please. This state of affairs is. The difference between public and
private is. No amount of hand waving will make the latter go away. All
you will accomplish by waving your hands is confusing people into posting
private material.
> Users must be aware that there
> are two protocols and two incompatible worlds at all times.
Of course, but whether those two worlds are called "news" and "email",
whether the protocols are "SMTP" and "NNTP", are both irrelevant. The two
"protocols" that users must be aware of are "private" and "public".
Making news and email look the same will blur the line between the two.
> Software must not
> smooth over these differences, and no attempt made to make these worlds
> compatible.
You have shown a remarkable inability to interpret simple english.
> It is here where many of us disagree.
Of course. You have created an argument that you intended to shoot down.
It's just too bad it's not my argument.
> I see absolutely no reason why I should have to care whether something is news
> or mail. It's up to these expensive computers to figure that detail out.
I don't care whether you care if your response is public or private, but
there are users who do care. There are admins who would rather not see
their archives of material from public places fill up with "how do I join
this group" or "Hey, Paul, call me at the office." messages. There are
readers of those groups who would rather their mailboxes or newsreaders
did not fill up with those sorts of messages.
With even the current glaring dichotomy between public and private, there
are still too many "how do I join this group" etc. messages. It can only
get worse when the division becomes less obvious.
> Ideally, there would be no Newsgroups: line. Newsgroups would appear on the
> To:, cc:, or bcc: lines where they belong.
So how do you handle the situation where a username and a newsgroup name
collide? I find it fascinating that a system where mail to some user who
is unlucky enough to have a username that is taken by a new newsgroup
would have private mail addressed to him posted to a public forum, is
something that you find desirable.
> But we live in an imperfect world.
> The actual integration effort is much more modest in its scope and much less
> disruptive.
In your opinion. But you have already said that you don't care about the
Leslie Mikesell
John Stanley <sta...@skyking.OCE.ORST.EDU> wrote:
>It ignores the private nature of much mail. Don't think that it is being
>ignored? Look at all the people here who are saying that mail is a
>public messaging system.
Correctly so.
>It doesn't matter if my US Mailbox is made out of aluminum or bronze, or
>whether the mail comes in the morning or the afternoon, or is brought by
>a woman or a man, or whether my mailbox is a post office box. The letter
>from Aunt Sally is still a private message.
Maybe it is, and maybe she sent copies to everyone she knows. If you
want to counter something she said, you deserve the opportunity to
reach the same audience. With email you sometimes have that capability.
>And if I can't get anyone to understand that the nature of the system is
>that way, how am I going to get them to remember that when they design
>the next UI?
Because you don't seem to understand it yourself. One solution is to simply
arrange to have more than one mailbox if you feel compelled to separate
your private messages from the rest, just as you might use a PO box or
an unlisted personal phone number. But, if you put your personal address
on a public message, you shouldn't be too surprised if people reply to it.
Les Mikesell
l...@mcs.com
Robert A. Rosenberg
wrote:
>> In a correct world, the MUA/NUA would have two functions:
>
>> 1) Reply personally to sender only
>> 2) Reply to sending medium (the list, or newsgroup) only
>
>> [ ignore cases of mixed medium messages for now ]
>
>I disagree. When I'm engaged in a serious discussion I want my replies to go
>directly to the people directly involved in the discussion. I also want a copy
>of my reply to go back to the list.
I agree with what you are saying (and the way you work). There are a number
of times when I reading a Newsgroup and want to send my message via Email
(permitted by my news agent as an alternative to doing a Reply to Newsgroup)
while at the same time posting it to the newsgroup for others to read. The
news agent presents me with an either/or situation where I must send
separate Email and Posting copies as opposed to being able to flag the post
to be send via an Email cc to someone (or, what is logically the same thing,
to post a copy of my Emailed reply to a newsgroup article back into the
newsgroup). There are often times when I send a direct copy of a Mailing
List Reply to someone as well as the list since, because the list is being
digested, this will short circuit the delay in delivering the message to the
intended party. Newsgroups also take time to propagate while sending a cc'ed
Email version gets there immediately.
I think that the best way of handling the question of Newsgroups and Email
is to have Headers in both the Newsgroup and Email version of a Dual Version
Message (ie: one that is both being Posted and Emailed by the News Agent)
which document the fact that the message/article is being transmitted by
both methods (ie: Document this fact). This allows the user to say where the
message is coming from and to control how replies are handled.
Gateways introduce an extra complication since there needs-to [/should] be
some indication in messages/articles that a mailing list and newsgroup are
linked by the Gateway so as to give the user on each end some control over
the Gateway Process. If I am replying to a message that started as a News
Post and was gated to an Email List then, when it is replied to by a List
recipient, the Email-to-News Gateway should be supplied in the Email Header
with the information to allow it to treat the News Article it is creating as
if it were a News Reply. This includes the capability to designate in an
Email Header field that the message should be cross-posted to other
newsgroups in addition to the newsgroup that is linked to the mailing list.
Thus if an Article is Crossposted to Newsgroups A and B (with A getting
gated to list C) when someone replies to that Email Message, the reply
should be crossposted by the gateway to both A and B and not just be gated
back to A.
John Stanley
James C Deikun <jcds...@pitt.edu> wrote:
>In article <38j5iq$k...@gaia.ucs.orst.edu>,
>John Stanley <sta...@skyking.OCE.ORST.EDU> wrote:
>>That's right, but it happened. But when there becomes a "standard" that
>>says that a "newsgroups" header means mail is supposed to be (or
>>already was) posted, you will be hard pressed to say that that same
>>transport is broken. "That header says it will be posted, so what's the
>>problem if I do it here?"
>
>There is no problem. Where else would the responsibility for posting
>it lie?
If I send mail to "usenet...@fred.com", the responsibility for
interpreting the username (mailbox) lies ONLY with the system
"fred.com". It is not the responsibility of "abc.edu", even if the
mail for fred.com passes through "abc.edu" on the way to fred.com.
And, in case you haven't been reading, someone suggested exactly such a
processing step: if a username looks like a newsgroup, post it right
away. Ignore the tiny detail that you aren't the host in the address.
>>No. Preventing things from breaking when someone creates a new meaning
>>for existing headers.
>
>We are, you aren't. You're trying to make sure no one creates a new
>meaning for existing headers.
Which will be a great inducement for things to break.
>You have offered absolutely no solutions
>to the problems that might occur if someone does,
Of course not. I am trying to get it not to happen. I don't have to
solve a problem that doesn't happen.
>yet you have made no convincing arguments that these problems are insoluble.
The problem will be insoluble if some idjit digs his heels in and
refuses to change the behaviour of his mailer, based on his feeling
that "of course you wanted it posted or you wouldn't have included the
header, and why is it a problem if my system posted it for you instead
of the system you sent it to?"
>You can only
>point to broken implementations. This shows absolutely nothing other
>than your pessimism and your ability to whine.
It shows beyond any doubt that the problem will occur -- a problem that
can be avoided by refusing to change the definition of headers already
in use.
Craig_E...@transarc.com
e-m.. John Gardiner Myers@cmu. (3191)
> > What if you send email to some list server somewhere asking it to
> > subscribe jg...@CMU.EDU to it? The list server might send its output
> > with the To: line showing some list...@somewhere.edu. How does the
> > software at CMU know to put it in a shared message base? If it
> > doesn't, have you really achieved any News/mail unification?
> What does the address in the To: line have to do with delivery? It's
> the *envelope* recipient addresses that matters.
Just to agree and emphasize: indeed, it's only the envelope address that
matters, and the original header lines won't usually even contain the
included jg...@CMU.EDU, or for that matter the
arpalists+h...@andrew.cmu.edu address. Local delivery of mail
to andrew.cmu.edu includes the step of prepending a Received: header
that captures the envelope address that was used in locating the given
local mailbox. Well, it might include the envelope address only for
local addresses that included some non-null text after the ``+''
separator; I forget. In any case, it's the envelope-destination address
that's important to capture, and it's therefore that envelope address
that is captured in a Received: header.
Craig
John Stanley
>The "whole camel" is already happily entrenched in the "tent." There
The "whole camel" is still outside the tent in a lot of places. The fact
that you might have let him in where you are is not proof that he should
be let in here.
>still has to learn to use two different programs, and, if he or she
>can't figure out how to set his or her editor, often ends up using
>two COMPLETELY different interfaces to perform the same function.
>I don't understand what kind of reasoning leads you to proclaim that
>this is LESS confusing than unification.
The use of two different interfaces is a constant reminder that there is
a difference. Remove the difference in the interfaces (i.e., make them
unified) and you remove the reminder.
>A good kludge for now would be to zap the Newsgroups: header or modify
>it to something on the order of Part-of-a-discussion-originally-on:
This kludge requires fixing all the software in use now. Not putting
extra meanings onto "Newsgroups" means writing NEW software differently.
Not "not writing new software", just writing it differently. There is
several orders of magnitude of difference between changing everything
already in use and writing new.
>>>and news, if not identical, at least mutually consistent.
>>
>>They are.
>
>The presence of a Newsgroups: header in a news article means the
>article was posted to the indicated newsgroups. Thanks to bad
>newsreaders, and as you repeatedly emphasize, in current practice
>the presence of a Newsgroups: header in mail currently means no
>such thing.
It means nothing. That is completely consistent: mail isn't news, so
"newsgroups" should have NO meaning in mail.
>This is obviously an inconsistency, if you use any
>sane brand of logic, and ought to be fixed.
The fact that some newsreaders leave "Newsgroups" in mail they send is
hardly any excuse for making is suddenly mean something. Especially
something as incorrect as "this needs to be posted to news". At
best, it means "this came from news", but only to people who read
headers. It means nothing to mail, and should continue to mean nothing,
simply because what the newsreaders putting it in mail now believe it means
and what news believes it means are two different things.
>>One current practice is to name the mailbox for a mail-news gateway
>>something like comp.mai...@some.site.com. I have had a helpful
>This is why I would like to see that role belonging to Newsgroups:
>instead of To:.
And that will cause exactly the problem I am trying to keep from
happening. Someone using his current newsreader, replying to someone by
mail, with the explicit intention of making a private reply, would find
his mail being dumped into news, if "newsgroups" in mail is suddenly
defined to mean the same thing as "newsgroups" in news.
>Since the Newsgroups: header would be a MESSAGE header rather than
>specifically a mail header, and since it's not used by SMTP, and
If it is in mail, and it is defined to mean something in mail, then it
will be a mail header.
>simply because of the semantics of it, the sender/posters site not
>only would be justified in handling the header, but would be
>unjustified in NOT handling the header if it has performed some
>sanity checks to make sure it is not including such headers as
>the result of someone's broken newsreader screwing up.
How do you program sendmail to know the difference between a broken
newsreader leaving "Newsgroups" in and someone putting it in
deliberately?
>>Of course, creating new headers that contain the information you want
>>(and, for the slow readers, changing software to honor and create those
>>headers) will not have such a problem.
>
>It would be preferable in creating a truly unified interface to make
>sure that at least the *internal* representations of mail and news
>are self-consistent and unified.
Defining news-specific headers in mail is not "self-consistent". Mail is
not news. It need not understand news-specific headers.
John Stanley
Ned Freed <N...@sigurd.INnosoft.COM> wrote:
>You must operate in the most tyrannical of working environments if you actually
>see things this way!
Why? Because I know that email is not the best medium for communicating
between two people? That fact would be true no matter where I worked.
>I work for a company where different employess work very different hours,
So do I. Sometimes face-to-face isn't practical. That is not,
however, proof that mail is a public medium; not even proof that mail is
better than face-to-face for communicating. It is certainly not proof
that every site in the world makes the same uses of mail that yours
does.
John Stanley
Mark Crispin <m...@panda.COM> wrote:
>Hi Rahul. I knew what was going on. I was flamed by an individual who
>lept to a completely unwarranted conclusion about my environment. I
No, Mark, you were flamed because you couldn't figure out how to stop
sending your replies to me both by mail and in news. I made no
conclusions about your environment. I shouldn't have to. Isn't that the
rallying cry of those who want to "harmonize etc."?
Here, though, is what I conclude about your environment, based on your
statements here, and from your headers: you use Pine, but you use a
version that is different from any version that I have ever seen, in
that you cannot edit or select the list of recipients for your mail.
You are a victim of Pine. It holds you hostage, forcing you to use it.
>how/what/why, one can not make the presumption that some other victim
>would know why he is receiving such abuse from this individual.
I told you why I used harsh language with you Mark. Very explicitely. Are
you trying to say, now, that you don't know why?
>It is *wrong* to flame someone for being at the other end of a dual
>environment discussion. It is *wrong* to flame someone because the
>deliver of his messages reflect the limitations of the medium. The key to
>the matter is to recognize this, and not to make vulgar accusations.
Mark, as someone who claims such great things for Pine, I would expect
you to be the last person to claim that you have limitations in how you
reply to something. Is that your claim, now? That you cannot select to
whom you reply when you send replies? That you are physically unable to
choose to reply only to "header-people"?
>> I'm sending this posting to comp.mail.headers, but no doubt I will soon
>> be accused of sending email to header-people.
>
>I doubt that anyone would accuse you of sending email to header-people,
Why not, Mark? You accused me of sending mail to header-people.
>because you don't send vulgar messages accusing people of stuffing your
>Very Important email inbox!
If you had honored a simple request when I made it, you would have
received nothing vulgar from me. When you decided to ignore someone
else's choice of medium for this discussion, you made your own bed.
When you made it clear that your choice to ignore that request was
because you were going to be stubborn, you tucked yourself in.
Mark Crispin
> Excerpts from internet.header-people: 23-Oct-94 Re: Harmonzing Internet
> e-m.. James C Dei...@pitt.edu (4360)
> > Mailers that leave Newsgroups: headers in mail in reply to news articles
> > are broken. Period.
> OK, maybe I'm just confused, but I don't know why you think they're
> broken.
Hi Craig. I think James' reasoning is due to this scenario.
1) User A posts a message to news.
2) User B sends a snotty reply in e-mail, leaving the Newsgroups: header
in the reply.
3) User A flames user B back using the reply-all function. His flame goes
to news as well as email, because user A's mailer (*and* user A the
human) thinks that user B posted the message.
James contends, correctly in my view, that the problem was caused by user
B's mailer falsely implying that user B posted to news as well; and that
this behavior should be considered broken.
The contention of certain other individuals is that this scenario means
that mailers must not have any form of reply function that includes the
contents of Newsgroups: headers in the reply. They argue that mailers
which falsely include Newsgroups: in mail headers will never be fixed
These individuals' proposal is to define some new header that duplicates
Newsgroups: but is guaranteed not to appear in messages that are not
posted to news. Until this new header is deployed, news-reply is
be permitted unless the program is in the mode of reading news.
These individuals turn a deaf ear to the argument that in many
environments it isn't possible to decide that a message is ``news'' as
opposed to ``mail'' -- and in fact argue that those environments are
poorly-designed and broken. Their model of a ``good'' design is one in
which mail and news are strictly separated.
Pine goes through the following steps when replying to a message:
1) ask ``Include original message in Reply?'' unless this is configured
by default. If YES, insert original message in reply text.
2) ask ``Post follow-up message to news group(s)?'' if the message has
a Newsgroups: header. If YES, set Newsgroups: of reply from
original message, go to step 5.
3) set To: of reply from primary reply address of original message.
4) ask ``Reply to all recipients?'' if there were additional recipients
of the original message. If YES, set cc of reply from original
message.
5) enter message composer, with headers displayed and ready for user to
enter the text.
The significant point is that the reply goes to news or to email based
upon the user's choice in step (2). It never goes to both. There is no
``reply-to-all-and-news'' function.
Step (2) enrages John Stanley, because it permits the user to post a reply
to email if he answers YES to this question in the scenario given above.
What is also likely to annoy him is the fact that once in the message
composer, it is possible to add both news *and* mail recipients in the
same message [considering his incorrect accusations that I was doing that].
I don't want a useful capability stripped, nor held hostage to the
deployment of some new header in news. I especially don't want this to
happen just because my having this useful capability annoys some fool who
goes on hearsay, can't be bothered to investigate, and who leaps to false
conclusions.
Mike Schenk
>Craig_E...@transarc.COM writes:
>> Excerpts from internet.header-people: 23-Oct-94 Re: Harmonzing Internet
>> e-m.. James C Dei...@pitt.edu (4360)
>>
>> > Mailers that leave Newsgroups: headers in mail in reply to news articles
>> > are broken. Period.
>>
>> OK, maybe I'm just confused, but I don't know why you think they're
>> broken.
>
>The mailers are broken because they are using the same header name for
>two completely different and incompatible semantic meanings.
In a mail message a Newsgroups: header does not have any semantic
meaning. Therefore, by definition it is not wrong to include a
Newsgroup: header in a mail message.
That is how mail is standardized. Now if you start coming up with your
own definitions according to your own misguided and retarded ideas, that's
your problem, but don't bother other people with it and don't
implemented anything to your ludicrous misconceptions!
Mike
Mike Schenk
>Even assuming Reply-To does work, there's no standard "Reply" method,
>and some MUA's will still insist on Cc'ing the original sender
>personally. (arguably they are broken, but they do exist)
How can one claim that an MUA is broken!? The standards only pertain to
interface (like any standard). So what your MUA adds to the headers
needs to be syntactically correct, that's all. Everything else is
thought up by people who are trying to expand the standards (RFC822 in
this case) to areas where they don't belong. This causes a problem
because everyone does this in a different manner.
Mike
halasz
> Perhaps this is a good example to debunk the foolish notion that email and
> have particpated solely as a member of header-people. Obviously, there is
> now a newsgroup that gateways header-people mail, and Stanley does not
> realize that when he posts to that newsgroup he also send mail to a wider
> audience.
>
lack of distinction, which arises solely from a natural but wholly
implementational likeness of email and netnews. The end user needs a strong
distinction: one who sends a letter expects particular behavior from its reder:
one sent to a friend is not further publishd, one sent to a company is publishd
only to those who fulfill its request (although it is a mistake for the writer
to assume that no:one else sees it), and one sent to a newspaper-editor may be
printed in the paper for all to see.
.....
> If you can admit your mistake, then you can start being part of the
> solution instead of continuing to be part of the problem.
Until you and other implementors make it trivially easy for a user to tell the
distribution of some email-address, you are more part of the problem than any
user who did that which Stanley did.
Rahul Dhesi
>Sorry--I should have said ``reply to readers'' rather than ``reply to
>recipients''. I don't think there's any difference between ``reply to
>readers'' and ``post follow-up''. That is, I think that your items 2
>and 3 are the same.
'Reply to readers' means one of two things.
1. Reply to those people who read the message to which you are replying.
2. Reply to those people who will read what you write.
In the case of a mailing list, the two sets of readers are likely
identical, so 'reply to readers' is not ambiguous. In the case of
Usenet, the two sets of readers are likely far from identical, so
some clarification of your terminlogy is needed.
--
Rahul Dhesi <dh...@rahul.net>
Craig_E...@transarc.com
e-m.. Keith Mo...@cs.utk.edu (971)
> What we need is a new header or two to indicate message transformations
> such as are caused by gateways and mailing list exploders. Right now
> all we have are Resent-* and Received, neither of which is sufficient.
I respectfully disagree. Received: is *exactly* the right header for
this, assuming that the message doesn't get out of the MTA environment
into the MUA environment, in which case ReSent-XXX are the right headers.
Of course, I also believe that the redistribution function shouldn't be
mucking with the message headers at all, except to prepend one or more
Received: lines.
Craig
Craig_E...@transarc.com
e-m.. John Sta...@skyking.oce (2195)
> And, in case you haven't been reading, someone suggested exactly such a
> processing step: if a username looks like a newsgroup, post it right
> away. Ignore the tiny detail that you aren't the host in the address.
I hope you weren't counting my posting, in which case I was talking
about mapping destination addresses to newsgroup names, but:
(a) only in the MUA, not the MTA
(b) only if the @domain was the local one
Craig
Craig_E...@transarc.com
e-m.. Rahul Dh...@rahul.net (1443)
> In <4ie34RiSM...@transarc.com> Craig_E...@transarc.COM writes:
> >My advice would be to abandon the mail/news distinction, and instead
> >push MUA developers to ensuring that users know the difference between
> >reply-to-sender and reply-to-all/reply-to-recipients.
> The trouble with this discussion is that there are sharp differences
> between mailing lists and Usenet newsgroups. ('Vive la Overlap' does
> not imply 'Le plus des changes, le plus des meme-choses', if you will
> pardon my French.) In the context of a Usenet newsgroup 'reply to
> sender' has some meaning (interpretation: send private reply to author
> of posting). But 'reply to recipients' has no meaning in this
> context.
> The correct choice, when discussing email vs News, is between 'send
> private reply to author' and 'post public follow-up'. A follow-up is
> never sent to any specific people. A follow-up, from the point of view
> of the person posting it, has no specific recipients. Anybody who
> chooses to read that follow-up -- and there is no predefined list of
> such people -- becomes a recipient. Usenet is far more dynamic than
> may appear at first sight.
> You might call this a trichotomy:
> 1. Send private response to original author.
> 2. Send public response to all recipients, if you can identify them.
> 3. Send public follow-up to various news spool directories, where
> countless unidentifiable people will have access to it.
Sorry--I should have said ``reply to readers'' rather than ``reply to
recipients''. I don't think there's any difference between ``reply to
readers'' and ``post follow-up''. That is, I think that your items 2
and 3 are the same.
That'll teach me to be more careful about terms. I think I've already
used the ``reply to readers'' idea here, and it was a thinko that I
would have said ``reply to recipients.'' Sorry!
Craig
John Stanley
Here's Mark, trying to speak for other people again.
>James contends, correctly in my view, that the problem was caused by user
>B's mailer falsely implying that user B posted to news as well; and that
>this behavior should be considered broken.
User B's mailer is saying that the article which was replied to started
in a newsgroup. The header was copied from the incoming article to the
outgoing mail. (This is incorrect behaviour, but that isn't the issue
right now.)
You can think the B's mailer is implying the article was posted, but
software doesn't imply. The only reason you would infer this incorrect
interpretation is if you don't understand what B's mailer is doing.
That is the problem with codifying the existance of a "Newsgroups"
header in mail to mean anything. The software uses it for one thing.
You want it to mean another.
>They argue that mailers
>which falsely include Newsgroups: in mail headers will never be fixed
I have not seen anyone making this argument.
>These individuals' proposal is to define some new header that duplicates
>Newsgroups: but is guaranteed not to appear in messages that are not
>posted to news. Until this new header is deployed, news-reply is
>be permitted unless the program is in the mode of reading news.
Since the Newsgroups header in mail is undefined and used by some
mailers to mean something different than what it means in news, the new
headers suggested do not duplicate "newsgroups". They have a completely
different meaning than the de-facto meaning "newsgroups" in mail has
now.
>These individuals turn a deaf ear to the argument that in many
>environments it isn't possible to decide that a message is ``news'' as
>opposed to ``mail'' -- and in fact argue that those environments are
>poorly-designed and broken.
They are. If you can't tell if a message is news (public) or mail
(private in many cases, public in some) because your "environment"
doesn't tell you, then your environment is broken.
>Their model of a ``good'' design is one in
>which mail and news are strictly separated.
I do wish you would learn that you do a poor job of making arguments
for other people and stop trying.
>Pine goes through the following steps when replying to a message:
Yummy. Too bad everyone in the world doesn't use mail the same way you
do, and doesn't use Pine to do it that way.
> 2) ask ``Post follow-up message to news group(s)?'' if the message has
> a Newsgroups: header. If YES, set Newsgroups: of reply from
> original message, go to step 5.
>Step (2) enrages John Stanley, because it permits the user to post a reply
>to email if he answers YES to this question in the scenario given above.
It does? I see nothing outrageous about step 2. It is correct behaviour
to include a newsgroups header if you are really replying by news. It
also doesn't mention mail in any way.
However, if you do not know whether the message you are replying to came
from news or not, then there is a problem. Am I enraged by this problem?
Hardly. Assuming a message came from news just because it has a
newsgroups header is blatantly incorrect behaviour, knowing, as you must
know by now, that some mailers leave that header intact in replies they
send my mail. Maybe you don't know this by now. Maybe you haven't
listened to that, either.
>What is also likely to annoy him is the fact that once in the message
>composer, it is possible to add both news *and* mail recipients in the
>same message [considering his incorrect accusations that I was doing that].
I never accused you of adding news recipients to anything. Get your
story straight, Mark. I accused you of refusing to remove me from your
mailed replies.
>I don't want a useful capability stripped, nor held hostage to the
>deployment of some new header in news.
(Sound of stomping feet...)
>I especially don't want this to
>happen just because my having this useful capability annoys some fool who
>goes on hearsay, can't be bothered to investigate, and who leaps to false
>conclusions.
What hearsay, Mark? I got the mail from you. I didn't have someone else
tell me I got it. Invesitgate what, Mark? How you received the article
you replied to? I don't need to. It doesn't matter. How you replied to
the article? Again, I don't need to investigate. The headers tell me.
The only false conclusion I had was that you were able to use your
system to control to whom you sent mail, and that you were smart enough
to realize that sending mail to "header...@some.address.somewhere"
would be sufficient to get your replies to me. Now I know better.
But, let's talk of "false conclusions". I find it hilarious that a
"header-people" expert wouldn't be smart enough to know the difference
between the "header-people" gateway sending him mail from the newsgroup
and my sending him mail directly. That is why you complained to my
postmaster about my sending you mail, isn't it, Mark, and why you
included news articles I had posted as proof that I was harassing you by
mail?
John Stanley
>In a mail message a Newsgroups: header does not have any semantic
>meaning. Therefore, by definition it is not wrong to include a
>Newsgroup: header in a mail message.
There is no formal definition of the "newsgroups" header in mail. That
is correct.
There is a current, de-facto definition which comes from those
newsreaders that leave the newsgroups header intact in mail they
generate in reply to news articles. This definition is: "the article to
which this is a reply originated in the following newsgroups".
There is some argument here that the definition should be "the article
you are currently reading has been posted to the following newsgroups:".
Steven D. Majewski
> There is no formal definition of the "newsgroups" header in mail. That
> is correct.
>
> There is a current, de-facto definition which comes from those
> newsreaders that leave the newsgroups header intact in mail they
> generate in reply to news articles. This definition is: "the article to
> which this is a reply originated in the following newsgroups".
>
> There is some argument here that the definition should be "the article
> you are currently reading has been posted to the following newsgroups:".
I would *almost* agree to the above as a reasonable summary.
What I would assert is that there is, in fact, no "current, de-facto"
definition. Some programs give "Newsgroups:" semantics, some give it
another. Judging by some of the comments made in this thread, and
looking at the headers my site and others generate, this must sometimes
be configurable, because I see (I think) the same programs generating
differnet headers. ( "I think" because I can't know what may be
caused by gateway header munging. )
John Stanley is correct that either interpretation is correct with
existing RFCs.
However, the second interpretation is in accord with Henry Spencer's
"son of rfc1036" (pre) draft. Admittedly - not even issued as a draft,
let alone an RFC. However, it's one of the best attempts we have to both
document existing practice and to set it straight where it doesn't
seem to follow a reasonable path.
And, for reasons argued endlessly here, I agree with that interpretation.
[ However, if Henry ever reads this thread, I'm afraid he'll give
up ever finishing and issueing that document as an impossible
and thankless task. Perhaps he already has run into this sort
of discussion, and that is the reason why is languishes in
pre-draft state. ]
-- Steve Majewski (804-982-0831) <sd...@Virginia.EDU> --
-- UVA Department of Molecular Physiology and Biological Physics --
-- Box 449 Health Science Center Charlottesville,VA 22908 --
John Stanley
Steven D. Majewski <sd...@virginia.EDU> wrote:
>I would *almost* agree to the above as a reasonable summary.
>What I would assert is that there is, in fact, no "current, de-facto"
>definition. Some programs give "Newsgroups:" semantics, some give it
>another.
The only semantics I have seen used in mail for the Newsgroups: header
is the one that means "I am responding by mail to a posting from this
newsgroup". I have never seen a mail message where it means "I posted
this to that newsgroup, too."
>However, the second interpretation is in accord with Henry Spencer's
>"son of rfc1036" (pre) draft. Admittedly - not even issued as a draft,
Son of RFC 1036 is a news RFC. How does it pretend to define mail
headers?
Michael Grubb
John Stanley <sta...@skyking.OCE.ORST.EDU> wrote:
>The only semantics I have seen used in mail for the Newsgroups: header
>is the one that means "I am responding by mail to a posting from this
>newsgroup". I have never seen a mail message where it means "I posted
>this to that newsgroup, too."
trn, for example, does the latter, when you enter a mail address in a Cc:
header in a followup. The Cc: header is stripped out and a copy is sent
by mail, with an intact Newsgroups: header, to the mail address specified.
I'd demonstrate this to you but....
-- M.
--
Michael Grubb <m...@ac.duke.edu> Duke University Academic Computing
phone +1 919 660 6903 / 127 North Building, Durham NC 27708-0134 USA
Barton E. Schaefer
} Subject: Re: Harmonzing Internet e-mail and Usenet News protocols
}
} In article <39trmr$e...@gaia.ucs.orst.edu>,
} John Stanley <sta...@skyking.OCE.ORST.EDU> wrote:
}
} >The only semantics I have seen used in mail for the Newsgroups: header
} >is the one that means "I am responding by mail to a posting from this
} >newsgroup". I have never seen a mail message where it means "I posted
} >this to that newsgroup, too."
}
} trn, for example, does the latter, when you enter a mail address in a Cc:
} header in a followup. The Cc: header is stripped out and a copy is sent
} by mail, with an intact Newsgroups: header, to the mail address specified.
Trn actually seems to be using BOTH semantics (eek). If you use reply-by-
mail instead of followup, it still leaves the Newsgroups: header, even
though the message has not been posted to any of those groups.
--
Bart Schaefer Vice President, Technology, Z-Code Software
scha...@z-code.com Division of Network Computing Devices, Inc.
civilization (siv"i-le-za'shen), n., see ISO 2022
Leslie Mikesell
Barton E. Schaefer <scha...@z-code.com> wrote:
>} trn, for example, does the latter, when you enter a mail address in a Cc:
>} header in a followup. The Cc: header is stripped out and a copy is sent
>} by mail, with an intact Newsgroups: header, to the mail address specified.
>Trn actually seems to be using BOTH semantics (eek). If you use reply-by-
>mail instead of followup, it still leaves the Newsgroups: header, even
>though the message has not been posted to any of those groups.
And for even more confusion, I sometimes want to respond to a message from
a different machine than where I read news, so I tell trn to "| mail ..."
and it arrives with the Newsgroups: header in place.
Les Mikesell
l...@mcs.com
Rahul Dhesi
E. Schaefer) writes:
>} trn, for example, does the latter, when you enter a mail address in a Cc:
>} header in a followup. The Cc: header is stripped out and a copy is sent
>} by mail, with an intact Newsgroups: header, to the mail address specified.
>Trn actually seems to be using BOTH semantics (eek). If you use reply-by-
>mail instead of followup, it still leaves the Newsgroups: header, even
>though the message has not been posted to any of those groups.
The cc: behavior is quite wrong, I think. It is mixing up the
semantics of email and News to treat the same thing as both a posting
and a message. However, the use of the Newsgroups: header is de facto
correct.
In a News posting Newsgroups: tells us in which newsgroups the posting
was intended to appear. This is traditional predictable behavior.
In email, the Newsgroups: header tells us that the person has sent a
private response to a posting found in the newsgroups in that header.
This too is traditional predictable behavior.
--
Rahul Dhesi <dh...@rahul.net>
Rahul Dhesi
>>...is broken. The CC header is undefined for news.
>What? How is it broken? RFC 1036 says nothing about how other
>headers (even 822) can't be used....
>_Therefore,_the_rule_is_adopted_that
>all_USENET_news_messages_must_be_formatted_as_valid_Internet_mail
>messages,_according_to_the_Internet_standard_RFC-822....
>However,_it_should_always_be_possible
>to_use_a_tool_expecting_an_Internet_message_to_process_a_news
>message.
A couple of sources of confusion here.
1. The original discussion of Cc: as used by trn was that the user
could give trn something with a Cc: in it, and trn would interpret it
as a request to send the posting by email to the designated recipient.
I don't think the final News posting retains the Cc: header -- somebody
please correct me if I am wrong. So, please let's be clear about what
we are discussing: the command language used between the user and trn
(which command language includes the Cc: header), or the News posting
that appears as a result of the command, which I don't believe includes
the Cc: header.
2. The (A = > B) => (B => A) fallacy.
The RFC excerpt says that it should be always possible to feed a valid
News posting to a program expecting email. It doesn't say that it
should be always possible to feed a valid email message to a program
expecting News. In fact it's easy to find counterexamples: Almost no
email message is valid as News, because most email mesages do not
contain a Newsgroups: header.
--
Rahul Dhesi <dh...@rahul.net>
Ned Freed
> happily delivered by almost all mail software. There is a requirement
> in the RFCs that a To: line be present, but that's not a requirement
> imposed by what my pro-Unification opponents call "email".
You never seem to get it, do you? Your limited personal experience simply does
not translate into a reasonable picture of what actually happens in the broader
world of email.
Requirements first. There is no requirement in RFC822 that a To: header be
present in all messages. The requirement is that at least one of To, Cc, and
Bcc must be present, and Bcc is allowed to be empty. There is an example of
this latter usage in section A.3.1.
And while it is true that many email systems will happily transport a message
with no To header, this practice is far from universal. There are several
fairly popular mailers that simply refuse to accept messages that don't conform
to RFC822 in terms of required headers. PP is one obvious example -- in fact it
is slightly more restrictive in the area of originator headers and insists on
message conformance to its own peculiar reading of RFC822, which I happen to
think is incorrect. (But that's another story.)
There is even one mailer I've seen which, when presented with a message that
has a bogus header, returns a series of 5xx status lines containing the text of
the section of RFC822 that has been violated. (Unfortunately its parser is
easily confused so it sometimes sends the wrong diatribe or even sends one when
there is in fact nothing wrong with the message.)
The actions that mailers that accept bogus header are also far from universal.
Approaches include:
(1) Generate a To: header from the envelope information.
(2) Generate an Apparently-to: header from the envelope information. (Note that
this doesn't make the message legal in format, so the problem then passes
down to some other mailer which can then reject the message.)
(3) Generate an empty Bcc: header.
(4) Send the message on but also return a copy to the originator saying that
the headers are bogus and need to be fixed.
There are also other approaches that are used in practice -- this is just a
sample.
As it happens news to email gateways are only one of the culprits that generate
messages with no recipient information in the header. The other common problem
is that someone sends a message specifying only Bcc recipients. Some mailers
take it upon themselves to delete the Bcc: header, even when doing makes the
header illegal. The message then fails at some remote point because of the
illegal header. Or, in a more amusing twist, some other mailer grabs the
message, says, "that header isn't legal", and applies either option (1) or (2)
from the above list, thus exposing the entire recipient list!
More generally, the fact that blind acceptance of malformed message headers is
not universal is best evidenced by the fact that many PC and Mac clients don't
generate a Date: header, which causes some mailers to reject their mail, which
then leads to tedious discussions on various lists. (The usual answer is that
the Mac Eudora user needs to tell the Mac their coordinates so the time zone
can be computed and a Date: header generated.) I see this happen with
depressing regularity at least once every couple of weeks. The fact that it
happens so often constitues proof that mailers are not in general as forgiving
as you might think.
There is also a school of thought that transport agents have no business
looking at headers anyway, and doing so in fact constitutes a layering
violation. The unfortunate fact of the matter is that given that some agents do
look and do reject messages, which means that other agents then have to start
looking and fixing in order to maximize interoperability.
Ned
Leslie Mikesell
Ned Freed <N...@innosoft.COM> wrote:
>There is also a school of thought that transport agents have no business
>looking at headers anyway, and doing so in fact constitutes a layering
>violation. The unfortunate fact of the matter is that given that some agents do
>look and do reject messages, which means that other agents then have to start
>looking and fixing in order to maximize interoperability.
I think we need at least two more rounds of integration (besides email/news)
before things are going to settle down. Netscape has already intgrated
an http browser interface with a newsreader and a mail sender. Now we
need a mail reader in there too so a single interface can do all the
interesting stuff. The other missing piece is to be able to read
local newsgroups from several sources or somehow emulate a local bbs
under an http reader (i.e. maintain a concept of who you are and what
you have seen in previous sessions).
Les Mikesell
l...@mcs.com
Ned Freed
> > not translate into a reasonable picture of what actually happens in the broader
> > world of email.
> I read this posting with interest. Once I skipped the name-calling I
> found some interesting information. However, it seemed unrelated to
> the original context of the (A => B) => (B => A) fallacy that led to
> this thread.
The reason for the posting wss quite simple: You seemed to think that the
email systems of the world don't enforce the requirements set forth in RFC822.
I responded by saying that this in fact is not the case.
Now as to what this has to do with anything, I really cannot say. All I noted
is that you used a couple of assertions to reach some conclusions, and at least
one of the assertions was unwarranted.
And far as name-calling goes, I call them as I see them. This is at least the
fifth time you've posted something containing a bunch of assertions and
conclusions where the assertions ran counter to reality.
Ned
Rahul Dhesi
>In article <3a9kvn$3...@hustle.rahul.net>, Rahul Dhesi <dh...@rahul.net> wrote:
>>expecting News. In fact it's easy to find counterexamples: Almost no
>>email message is valid as News, because most email mesages do not
>>contain a Newsgroups: header.
>It's easy to find a counter-counter example. No News message is valid
>as e-mail, since it doesn't have a "To:" line.
Treading-on-thin-ice warning!
Since the dawn of the email era, email without a To: line has been
happily delivered by almost all mail software. There is a requirement
in the RFCs that a To: line be present, but that's not a requirement
imposed by what my pro-Unification opponents call "email".
However, Usenet postings without a Newsgroups: header have never
been happily transported by Usenet software.
--
Rahul Dhesi <dh...@rahul.net>
Kari E. Hurtta
»In <3ab03a$c...@kronos.fmi.fi> hur...@dionysos.fmi.fi (Kari E. Hurtta) writes:
»>No, you can't feed a valid News posting to a program excepting email.
»Can I can successfuly feed a News posting to the command
»'sendmail hur...@dionysos.fmi.fi' and have it delivered?
»If so, the claim is proven to be false.
Hmm. Perhaps I didn't phrase my sentence carefully enough. But I think that
you get a point.
--
- Kari E. Hurtta / Elämä on monimutkaista
Kari....@Fmi.FI puh. (90) 1929 658
{hurtta,root,Postmaster}@dionysos.fmi.fi
Rahul Dhesi
>I think we need at least two more rounds of integration (besides email/news)
>before things are going to settle down.
You have chosen a good word 'integration'. I am reminded of the era of
integrated software, which combined things like spreadsheet, word
processor, and database. These were presented in a user interface that
allowed them to smoothly coexist. I am glad that nobody claimed that
integrating a word processor and a spreadsheet make them the same
thing, and I hope people will stop claiming today that integrating News
and email makes them the same thing.
--
Rahul Dhesi <dh...@rahul.net>
Rahul Dhesi
>No, you can't feed a valid News posting to a program excepting email.
Can I can successfuly feed a News posting to the command
'sendmail hur...@dionysos.fmi.fi' and have it delivered?
If so, the claim is proven to be false.
--
Rahul Dhesi <dh...@rahul.net>
Brian Edmonds
>> In other words, a message with a Cc: line must be processed by the
>> newsreader.
John> Of course, that is not an issue. But RFC 1036 does NOT say, as you
John> imply, that it MUST process the CC header. It should NOT, since CC
John> is NOT a defined header in news.
Actually, all RFC1036 defines is how the *transport* mechanism is to
deal with headers. What a NUA does with headers is entirely up to the
author of the program, so if trn wants to honour Cc headers, that's just
fine by RFC1036. This lack of definition of proper NUA behaviour was
the main motivator for the thread a while back on developing a "good
USENET housekeeping" certification for newsreading software.
------DP-----BGC-----Akira-----UF-----Project-Ako------/================
Brian Edmonds (Prodigy Technologies Corporation) / SCA: Bardas,OFLT
http://www.cs.ubc.ca/spider/edmonds ...munitions / Lions Gate, AnTir
----Depeche-Mode---Robert-Plant---Bruce-Cockburn----/___inux spoken here
http://sailfish.peregrine.com/wb/ww/m(92,42,0,0)
John Stanley
>trn, for example, does the latter, when you enter a mail address in a Cc:
>header in a followup.
Trn does the former when you make a reply by mail.
If trn actually sends mail when you use the followup command, it
Leslie Mikesell
I don't agree at all. The first generation of 'integrated software'
was just a crude interface layer over the old basic components and
as you remember, nobody liked it much mostly because the user still
had to be aware of the underlying functional differences. More recent
(and more popular) software hides the differences to a point where
the user only sees one thing. For example WordPerfect 6.0 combines
the WP function of tables with nearly everything older spreadsheets
can do (including reading their files), and has the ability to access
several types of databases directly for its mail-merge data. So, is it
still one thing or is it several? Users only want to see one.
In the case of a combined email and news interface, the integration
may be complete on the user interface side, in that a single program
on a client machine can connect to servers of each type. The only
difference at that end is the packaging of the packets for different
protocols to the servers.
Les Mikesell
l...@mcs.com
Ned Freed
> > before things are going to settle down.
> You have chosen a good word 'integration'. I am reminded of the era of
> integrated software, which combined things like spreadsheet, word
> processor, and database. These were presented in a user interface that
> allowed them to smoothly coexist. I am glad that nobody claimed that
> integrating a word processor and a spreadsheet make them the same
> thing, and I hope people will stop claiming today that integrating News
> and email makes them the same thing.
Strawman argument here, I'm afraid. I don't recall anyone on this list assering
that news and email are the same thing, or that integrating them will make them
the same. I certainly do not believe them to be the same thing, nor does
integrating them make them the same in practice.
For that matter, I don't believe the multitude of different types of email are
the same, nor are the many different types of news services the same. There are
substantial differences even within the individual categories, and often as not
an extremely delicate balancing act is required to balance the needs of
different communities using different and somewhat incompatible services.
Anyone who has spent any time at all looking at LAN email or X.400 integration
issues on the email side, or Notes integration issues on the news side, can
attest tto this.
I do believe, however, that users simply don't care about most of the
differences. All they want is something that works and does the right things at
the right time. This in turn implies that integrated interfaces have to go to
some lengths to preserve the real differences between the various services
behind the scenes while simultaneously not annoying users with gratitious
differences they care next to nothing about.
Users are therefore demanding integrated interfaces. And vendors have built
them and are building lots more. This trend is so prevalent that pretty soon I
suspect they will assume a dominant position in terms of actual usage. As such,
it is less a question of whether or not integration will occur but whether or
not we have the necessary understanding to make it all work properly.
Finally, while endless discussion of the abstract characteristics of various
systems is amusing, if only in how it highlight the limits of personal
experience, I still remain to be convinced that it provides any useful
insights here. If there's anything at all for this group to do it is to deal
with specific issues that arise when things are integrated. The recent posting
about the handling of references headers is a step in the right direction,
but it is just one step.
Ned
Rahul Dhesi
>I don't recall anyone on this list assering
>that news and email are the same thing, or that integrating them will make them
>the same. I certainly do not believe them to be the same thing, nor does
>integrating them make them the same in practice.
If by 'this list' you are referring to everybody who has been posting
here, and if you speak for all of them in saying that nobody is asserting
that News and email are the same thing or that unification will make
them the same thing, then this discussion/argument/flame war is now
over. We all agree.
Only if a header-person (singular for header-people) now posts a
contradiction will the flame war be back on again.
--
Rahul Dhesi <dh...@rahul.net>
Kari E. Hurtta
»2. The (A = > B) => (B => A) fallacy.
»The RFC excerpt says that it should be always possible to feed a valid
»News posting to a program expecting email. It doesn't say that it
»should be always possible to feed a valid email message to a program
»expecting News. In fact it's easy to find counterexamples: Almost no
»email message is valid as News, because most email mesages do not
»contain a Newsgroups: header.
No, you can't feed a valid News posting to a program excepting email.
Lets look RFC822's appendix A.3.1:
A.3.1. Minimum required
Date: 26 Aug 76 1429 EDT Date: 26 Aug 76 1429 EDT
From: Jo...@Registry.Org or From: Jo...@Registry.Org
Bcc: To: Sm...@Registry.Org
Note that the "Bcc" field may be empty, while the "To" field
is required to have at least one address.
How many news article have Bcc or To -header?
Robert A. Rosenberg
<Pine.A32.3.90.941107...@elvis.med.Virginia.EDU>,
sd...@virginia.EDU (Steven D. Majewski) wrote:
>However, the second interpretation is in accord with Henry Spencer's
>"son of rfc1036" (pre) draft. Admittedly - not even issued as a draft,
>let alone an RFC. However, it's one of the best attempts we have to both
>document existing practice and to set it straight where it doesn't
>seem to follow a reasonable path.
>
>And, for reasons argued endlessly here, I agree with that interpretation.
>
>[ However, if Henry ever reads this thread, I'm afraid he'll give
> up ever finishing and issueing that document as an impossible
> and thankless task. Perhaps he already has run into this sort
> of discussion, and that is the reason why is languishes in
> pre-draft state. ]
Once solution to the problem is to devalue Newsgroups in Son-of-RFC1036 and
replace it with 2 Headers. Once to designate the Newsgroups the Message was
posted in (Newsgroups-Posted-In) and one to designate the Newsgroups the
Reply is being posted to (Newsgroups-Posted-To). There should also be a 3rd
to control Follow-Ups. This last two is to assist Mail-to-News Gateways in
posting the article version of the message. I do not see any reason why the
Newsgroups-Posted-In Header is needed (since no equivalent exists in News)
but having it should keep those who use Newsgroups for this purpose happy
while allowing the "this reply was posted in" usage to also be supported.
Robert A. Rosenberg
>The cc: behavior is quite wrong, I think. It is mixing up the
>semantics of email and News to treat the same thing as both a posting
>and a message.
WHY? If I can reply to a message either via doing a POST or a
REPLY-VIA-EMAIL, why do you feel that adding a "Cc" header to the Article
when I Post it to tell the News Agent to treat the Article as if it were
being submitted via the REPLY-VIA-EMAIL Option as well as going through the
Post Path is wrong? This saves the need to create two copies of the message.
What headers get placed in the Actual Posted and Mailed Copies is a separate
issue from having one copy of a message act as both an Article to be Posted
and an Email Message. There is also a need for Headers to support
Synchronization of Gatewayed News and Mailing Lists. Having Headers that
document this requirement makes life much easier. The actual types a
information needed is a separate topic.
Robert A. Rosenberg
(John Stanley) wrote:
>If trn actually sends mail when you use the followup command, it
>is broken. The CC header is undefined for news.
The CC Header is a request to the news agent to treat the message as if it
were being sent via the "reply via Email" option. Its use is to eliminate
the need to send the same text as two separate options - Once as a Post and
once as a Reply.
Robert A. Rosenberg
Hurtta) wrote:
> A.3.1. Minimum required
>
> Date: 26 Aug 76 1429 EDT Date: 26 Aug 76 1429 EDT
> From: Jo...@Registry.Org or From: Jo...@Registry.Org
> Bcc: To: Sm...@Registry.Org
>
> Note that the "Bcc" field may be empty, while the "To" field
> is required to have at least one address.
Not Exactly - For the EMAIL to be valid, there must be an address in one (or
more) of the following 3 Headers - To, Cc, and Bcc. It is perfectly OK to
use the Cc field to address a message and leave the To empty (I've seen this
done when the main list is in Bcc and you want to openly send a Cc copy to
someone [who would be being Cc'ed even if the main address was in the To
Header or to avoid the Apparently-To Header that using only a Bcc causes]).
Robert A. Rosenberg
>I think we need at least two more rounds of integration (besides email/news)
>before things are going to settle down. Netscape has already intgrated
>an http browser interface with a newsreader and a mail sender. Now we
>need a mail reader in there too so a single interface can do all the
>interesting stuff. The other missing piece is to be able to read
>local newsgroups from several sources or somehow emulate a local bbs
>under an http reader (i.e. maintain a concept of who you are and what
>you have seen in previous sessions).
So long as we have the need/ability to link a Newsgroup with a Mailing List
(via News-to-Email and Email-to-News Gateways) there needs to be defined
Headers in both Email and News to control the routing of the Gatewayed
Messages/Articles. The Simplest way is to allow the Current News Headers to
be acceptable to Email (or at least the Gateway Code) as well as allowing
To/Cc/Bcc to be accepted by News (possibly short circuiting a News-to-Email
Gateway which should/would ignore any messages that lists its newsgroup
since the message has already being inserted into Email via the News Agent
doing a Reply-via-Email). To make the linking of a Mailing List and a
Newsgroup seamless there must be information in both the versions so
messages posted from either side can be mapped as they pass the gateway so
as to act as if they came from the far side of the gateway. Thus a Message
that starts as news must preserve its Newsgroup and Followup-To Headers as
is goes to the Mailing List so that if/when a reply is made, the message
gets correctly cross-posted to all newsgroups as it passes back through a
gateway (or gets handed off to a News Agent by the Mailing List Handler
Agent Software).