🚀 go-pugleaf

Unfortunately, you've hit on a problem: not all gateways are created
equal, and neither are all news-admins.

PSU NetNews inserts a Gated-By: header to indicate it came from a
mailing list.  All of the bit.* hierarchies are bidirectional, to my
knowledge, and Jim McIntosh (jim@American.edu) periodically posts infor-
mation about the gateways.

The newsgate software that Rich Salz wrote (it's in the alt.sources
archives - get the newer version, though) works just fine for C-News,
and can support INN as well.  I dunno about B-News.  We're using it
for a few local gateways here, and I can provide information if needed.

But there are others - mostly outside of bit.* - that do all kinds of
bizarre contortions.  Sometimes a system-generated header is put on
and the original header makes up part of the body of the article (try
replying to one of those right).

Others are "inet" newsgroups, which are newsgroups that started out
as mailing lists (just bring up the inet groups in news.admin and see
the howls...).  These usually go from the mailing list back to news
via ucbvax.berkeley.edu gateway.  I'm not sure if anyone but Erik Fair
understands these...

Then there's the broken ones... botched headers, headers that should be
filtered out (like Reply-To: pointed back to a mailing list), mail-news
loops, one-way gateways... unless they're in bit.*, or managed by some-
one that knows what they're doing, there's not much that can be done
about them but complain to the gateway manager.
--
Doug Sewell, Tech Support, Computer Center, Youngstown State University
doug@cc.ysu.edu	   doug@ysub.bitnet	<internet>!cc.ysu.edu!doug
"Recession: your friend loses his job.  Depression: you lose your job.
Recovery: Bush loses his job" - Bill Davidsen

Frankly, I spend far too much of my time REMOVING header material.
I wouldn't want any more.

A.
-- 
  Arthur Radford, Bristol Polytechnic    -------------------------------
  BRISTOL, BS16 1QY, UK                	      Esperanto parolata
  as_radfo@uk.ac.brispoly.csd            -------------------------------

In article <1992Jul15.151734.22118@csd.brispoly.ac.uk> as_radfo@csd.brispoly.ac.uk (AS Radford) writes:
>
>Frankly, I spend far too much of my time REMOVING header material.
>I wouldn't want any more.
>
>A.
>-- 
>  Arthur Radford, Bristol Polytechnic    -------------------------------
>  BRISTOL, BS16 1QY, UK                	      Esperanto parolata
>  as_radfo@uk.ac.brispoly.csd            -------------------------------

Well, the headers may be time consuming with some 1200 or 2400bps
connection, but at least with 9600bps connection, most of my time
goes for reading the couple of first lines of each article, and deciding
whether it is interesting or not. My eyes can automatically skip the
headers, that's not the problem, unless it is page or two long.

BTW. Obviously I wasn't thinking 100% when I posted that suggestion.
Of course we don't need no any additional Original-Sender field in header,
because that information is already in From: field. The problem is that
when I reply with R command, and news software takes the mailing list
address from the Reply-To: field into To: field, I have nowhere the
original address of the sender, if he hasn't included it in his signature.
  So maybe there should be another reply command available in newsreaders,
which would always use the address found from From: field, even if there's
another one in Reply-To: field. (For replying privately to the articles
gatewayed from mailing lists to newsgroups.)

Anyway, I think it's important that when people set up the gateways, they
inform about it to newsgroup users, and make clear whether the gateway is
uni- or bidirectional.
  But as the Usenet has no memory, and those gateways can still be working
after years, when the readers of the group may have largely changed, it
would be better if this informing would be done in some permanent way.
  So I still suggest using the header for that. If we could get some
"standardized" gateway-header, the information could be contained
in one line, and the finesses of the reply and follow-up commands could
be then explained in FAQ files.
So just something like:

Gatewayed-from: foobar-l (unidirectional)

And if you really dislike additional header fields, you can always modify
your newsreader to suppress them. It's much harder for the beginners to
modify newsreaders other way.

Hmm, with newsgroups solely created for gatewaying the mailing lists to
the Usenet (like various bit.listserv.* groups) would it be good idea
to make that group "pseudo-moderated", i.e. set the moderators address
to be the same as mailing list address, so every follow-up would surely
end to list. (Which itself could be moderated or not.)
  Maybe this is already done with some gateway lists, I don't know.

-- 
Antti Karttunen   <>   karttu@mits.mdata.fi   <>   Mopsi s|i mopin!

In article <1992Jul16.005647.10015@mits.mdata.fi> karttu@mits.mdata.fi (Antti Karttunen) writes:
>
>BTW. Obviously I wasn't thinking 100% when I posted that suggestion.
>Of course we don't need no any additional Original-Sender field in header,
>because that information is already in From: field. The problem is that
>when I reply with R command, and news software takes the mailing list
>address from the Reply-To: field into To: field, I have nowhere the
>original address of the sender, if he hasn't included it in his signature.

  Complain to the mailing list manager.  Complain to people who write
broken software which does this.  The "Reply-To:" is supposed to be
an optional header available to the author of the article.  The practice
of adding a "Reply-To:" pointing back to the list is abusive.  It abuses
the intentions of the "Reply-To:" header.  It abuses the readers who
wish to reply privately to the author.  It abuses the readers of the
mailing list who see many messages of no interest to anyone but the
original author.

In article <1992Jul16.023706.32373@mp.cs.niu.edu>, rickert@mp.cs.niu.edu (Neil Rickert) writes:
|> In article <1992Jul16.005647.10015@mits.mdata.fi> karttu@mits.mdata.fi (Antti Karttunen) writes:
|> >
|> >BTW. Obviously I wasn't thinking 100% when I posted that suggestion.
|> >Of course we don't need no any additional Original-Sender field in header,
|> >because that information is already in From: field. The problem is that
|> >when I reply with R command, and news software takes the mailing list
|> >address from the Reply-To: field into To: field, I have nowhere the
|> >original address of the sender, if he hasn't included it in his signature.
|> 
|>   Complain to the mailing list manager.  Complain to people who write
|> broken software which does this.  The "Reply-To:" is supposed to be
|> an optional header available to the author of the article.  The practice
|> of adding a "Reply-To:" pointing back to the list is abusive.  It abuses
|> the intentions of the "Reply-To:" header.  It abuses the readers who
|> wish to reply privately to the author.  It abuses the readers of the
|> mailing list who see many messages of no interest to anyone but the
|> original author.
|> 

Most lists are set such that any Reply-To inserted by the author will be
respected; the list is only inserted if there is no Reply-To. So, this is
under the poster's control (though how a news gateway behaves regarding
Reply-To is not a function of poster or LISTSERV). Perhaps the gateway should
insert the From: address into Reply-To.

If these Reply-To were so abusive, why have people been using these lists for
so many years? Obviously, people do not feel abused. I don't see your name on
RFC 822, so what makes you so certain of the intentions of Reply-To?

This boils down to the BITNET camp vs. Internet camp again, which is a rather
old debate. Nobody asked Internet users to use LISTSERV...if they don't like
it, they're free to stop using it or write a better product (which nobody has
done). LISTSERV wasn't designed to gateway into news, so it is not LISTSERV's
responsibility to change just so people can use it for this. How a gateway
converts the mail into a news article is the gateway's concern. What headers
the user can access or use for a reply is the user agent's concern.

The big debate on ietf-list-wg, before many of us gave up, realizing that a
consensus was impossible, was whether the new mailing list manager (or
whatever you want to call it) generated a new mail message or simply
distributed the mail message. LISTSERV currently operates by generating a new
mail message, so it is free to put whatever headers it pleases into the
message. I remain in the "new message" camp...otherwise all you would have is
a glorified alias expander and alias list manager. Perhaps this is what some
people want.

The contents of Reply-To are controlled by the poster; the list owner selects
the default. The problem the original poster is having is, I agree, an
undesirable situation; it is not LISTSERV's problem, though. Complaining to
the list owner is a reasonable action; calling LISTSERV "broken" shows an
unjustified arrogance.

--Shane

In article <1992Jul16.005647.10015@mits.mdata.fi>, karttu@mits.mdata.fi (Antti
Karttunen) says:
>Of course we don't need no any additional Original-Sender field in header,
>because that information is already in From: field. The problem is that
>when I reply with R command, and news software takes the mailing list
>address from the Reply-To: field into To: field, I have nowhere the
>original address of the sender, if he hasn't included it in his signature.

Posts coming from a Listserv mailing list (such as those in the bit.**
newsgroups) should have a valid From: address, as well as a Reply-To:
address which is configured at the option of the list owner. If the list
owner prefers that replies be directed back to the original poster then
they specify that in the list definition and the Reply-To: field will
contain the e-mail address of the poster. If the list owner prefers any
follow-up discussion appear on the list, then the list is configured so
that the Reply-To: field points back to the list address.

People who receive the posts through the mailing list have the option,
then, of writing directly to the sender or to the list, or accepting the
preference of the list owner regarding additional discussions and
honoring the Reply-To: address.

I can see your point that for people reading these same posts in a
newsgroup, the option is to either reply publically by posting a response
to the newsgroup or replying privately by writing a piece of e-mail. The
Reply-To: generated by the mailing list, although reflecting the
perferences of the list owner, really only serves to get in the way of
the Usenet reader.

>Anyway, I think it's important that when people set up the gateways, they
>inform about it to newsgroup users, and make clear whether the gateway is
>uni- or bidirectional.

I agree completely. Gateways to and from the bit.** newsgroups, as well
as any additional gateways I've set up here (ROOTS-L to soc.roots, for
example) have only been established following the completion of a
procedure which calls for proper notification of the list owner, the list
subscribers and the newsgroup readers. Only if there are no substantial
objections are the gateways established.

An additional point: there are no uni-directional gateways in the bit.**
groups. When people have asked about them I've strongly discouraged them
and suggested moderated newsgroups as a prefered solution. The biggest
concern on the part of the list owner and subscribers is that Usenet
readers have been known to, um, er, generate inappropriate posts at
times, and the mailing people want to protect their list as much as
possible. Moderated newsgroups allow the list owner to intercept the
inappropriate while still allowing the Usenet readers to contribute to
the ongoing discussions.

>  So I still suggest using the header for that. If we could get some
>"standardized" gateway-header, the information could be contained
>in one line, and the finesses of the reply and follow-up commands could
>be then explained in FAQ files.
>So just something like:
>
>Gatewayed-from: foobar-l (unidirectional)

Any posts gated by the Penn State NetNews package (which is the one we
run here to gate the bit.** newsgroups) inserts a line saying that the
post came through a gateway. I've seen similar lines in posts gated at
Berkeley. I would think that any well behaved gateway should do the same.

>Hmm, with newsgroups solely created for gatewaying the mailing lists to
>the Usenet (like various bit.listserv.* groups) would it be good idea
>to make that group "pseudo-moderated", i.e. set the moderators address
>to be the same as mailing list address, so every follow-up would surely
>end to list. (Which itself could be moderated or not.)
>  Maybe this is already done with some gateway lists, I don't know.

Any posts coming from a Listserv list are posted to the appropriate news
group (ie, approved tags are inserted). When the mailing list is
unmoderated, the corresponding newsgroup is unmoderated and any posts are
sent by the gateway site to the mailing list. When the mailing list is
moderated or the list owner has a concern about Usenet rowdiness, the
newsgroup is moderated and the moderator address points to the list
owner.
---
Jim McIntosh (jim@american.edu)
The American University
Washington DC 20016-8019 USA

In <1992Jul16.165608.17970@utdallas.edu> shane@utdallas.edu (Shane Davis) writes:

>In article <1992Jul16.023706.32373@mp.cs.niu.edu>, rickert@mp.cs.niu.edu (Neil Rickert) writes:
>|>   Complain to the mailing list manager.  Complain to people who write
>|> broken software which does this.  The "Reply-To:" is supposed to be
>|> an optional header available to the author of the article.  The practice
>|> of adding a "Reply-To:" pointing back to the list is abusive.  It abuses
>|> the intentions of the "Reply-To:" header.  It abuses the readers who
>|> wish to reply privately to the author.  It abuses the readers of the
>|> mailing list who see many messages of no interest to anyone but the
>|> original author.
>|> 

>Most lists are set such that any Reply-To inserted by the author will be
>respected; the list is only inserted if there is no Reply-To. So, this is
>under the poster's control (though how a news gateway behaves regarding
>Reply-To is not a function of poster or LISTSERV). Perhaps the gateway should
>insert the From: address into Reply-To.

>If these Reply-To were so abusive, why have people been using these lists for
>so many years? Obviously, people do not feel abused. I don't see your name on
>RFC 822, so what makes you so certain of the intentions of Reply-To?

I certainly feel abused by this practise.  I had to write a script to bounce
undeliverable mail (account disk space full, no mail folder, etc.) on our
mailframe.  Messages from a LISTSERV can't be handled in the same way as
regular mail because the Reply-To: header line points back to the list -- 
bouncing the message to that address causes a mail loop.  The only 
solution seemed to be to check the comments in header lines for tokens
like "list" or "distribution" that are generally found in messages from
a LISTSERV.  I'd be a lot happier though if LISTSERV would add an
Errors-To: header to provide a proper place to report delivery problems to.

I'd guess that LISTSERV also has some klugey logic to try to distinguish
a bounce reply from a list submission.  
-- 
John Stewart -- Computing and Communications Services, Carleton University
Internet: jstewart@ccs.carleton.ca or Mr.Canoehead@algonquin.carleton.ca  
"Why can't we just forget about our constitutional problems and go canoeing?"

In article <jstewart.711310628@cunews>, jstewart@alfred.carleton.ca (John Stewart) writes:
|> 
|> I certainly feel abused by this practise.  I had to write a script to bounce
|> undeliverable mail (account disk space full, no mail folder, etc.) on our
|> mailframe.  Messages from a LISTSERV can't be handled in the same way as
|> regular mail because the Reply-To: header line points back to the list -- 
|> bouncing the message to that address causes a mail loop.  The only 
|> solution seemed to be to check the comments in header lines for tokens
|> like "list" or "distribution" that are generally found in messages from
|> a LISTSERV.  I'd be a lot happier though if LISTSERV would add an
|> Errors-To: header to provide a proper place to report delivery problems to.
|> 
|> I'd guess that LISTSERV also has some klugey logic to try to distinguish
|> a bounce reply from a list submission.  

LISTSERV catches bounces (1) if "Undeliver" is in the Subject: (2) if the
sender is one of a set of typical error senders such as "mailer-daemon",
"postmaster", "root". Yes, kludgey.

An Errors-To: header is the right way to go; I can't say why it hasn't been
done.

Where is the right followup for this? I removed n.s.readers from this one.

--Shane