From geoff@world.std.com Mon Oct  7 15:44:10 1991
Received: from BBN.COM by pineapple.bbn.com id <AA18404@pineapple.bbn.com>; Mon, 7 Oct 91 15:44:03 -0400
Received: from relay2.UU.NET by BBN.COM id aa06563; 7 Oct 91 15:43 EDT
Received: from world.std.com by relay2.UU.NET with SMTP 
	(5.61/UUNET-internet-primary) id AA25897; Mon, 7 Oct 91 15:43:08 -0400
Received: from localhost by world.std.com (5.61+++/Spike-2.0)
	id AA10919; Mon, 7 Oct 91 15:42:57 -0400
Message-Id: <9110071942.AA10919@world.std.com>
To: Rich Salz <rsalz@BBN.COM>
Subject: Re: How do you set up a news/mail gateway with C news? 
In-Reply-To: Your message of "Mon, 07 Oct 91 15:23:33 EDT."
             <9110071923.AA05877@litchi.bbn.com> 
Date: Mon, 07 Oct 91 15:42:45 -0400
From: geoff@world.std.com
Status: RO

My basic complaint with all the gateways I've seen (and I actually
haven't read much of yours, but we do run it here) is that they go beyond
necessary header munging and delete headers like To: and Cc:.  One can
argue that if everything is set up correctly, that this is harmless and
presents the illusion that this stuff is all news and not mail, but it
breaks down when the mailpaths file isn't current (e.g. here at world).

I had to ask around to find out where the ietf-smtp (or ietf-822) list
was homed when I wanted to post a message to info.ietf.smtp (or whatever)
because (a) mailpaths here punts to uunet and it wasn't clear that uunet
did the right thing with info.all (b) the addresses in the mail headers
had been deleted.  In hindsight, it might have all worked if I'd just
posted, but it's not hard to imagine a case where a locally-gatewayed
group doesn't have a matching mailpaths entry.

Should the article get gatewayed back into mail, *then* the old mail
headers should get deleted, but probably not before.

Spike also comments that we seem to quietly lose articles and I noticed
that your code does use unchecked static buffers (for shame!).  I think
the libraries that come with C News make it fairly painless and cheap
(though a few routines are due to be tuned further) to just use dynamic
strings everywhere, though I'm working on a more general dynamic strings
library.  The C News libraries are useful enough that lots of people use
them for random applications and we're thinking of posting them
separately from C News.

From per@erix.ericsson.se Tue Dec 17 15:24:13 1991
Received: from BBN.COM by pineapple.bbn.com id <AA23290@pineapple.bbn.com>; Tue, 17 Dec 91 15:24:08 -0500
Received: from mailgate.ericsson.se by BBN.COM id aa03729; 17 Dec 91 15:18 EST
Received: from super.eua.ericsson.se by mailgate.ericsson.se (4.1/SMI-4.1-MAILGATE1.7)
	id AA05327; Tue, 17 Dec 91 21:18:26 +0100
Received: by super.eua.ericsson.se (4.1/SMI-4.1.2)
	id AA24442; Tue, 17 Dec 91 21:18:25 +0100
Date: Tue, 17 Dec 91 21:18:25 +0100
From: Per Hedeland <per@erix.ericsson.se>
Message-Id: <9112172018.AA24442@super.eua.ericsson.se>
To: rsalz
Subject: Re: mail2news leaves files in /tmp
Newsgroups: news.software.b,comp.mail.misc
In-Reply-To: <3975@litchi.bbn.com>
References: <1991Dec04.122651.25792@csource.oz.au> <1991Dec14.133643.1284@eua.ericsson.se> <1991Dec15.153055.22201@mp.cs.niu.edu> <1991Dec15.191054.8154@eua.ericsson.se>
Status: RO

In article <3975@litchi.bbn.com> you write:
>I now think mail2news should do all the header-modifying and just pass the
>articles on to whatever program you like.  This means supporting some common
>subset of inews flags.

Yes, that's very nice, however as Neil Rickert kindly pointed out, -o is
already handled by mail2news; I don't need -x (of course I knew that,
more below...); and -= avoids the -h; - which (along with -d, -a etc)
means just about everything one needs is already in mail2news (this is
version 1.6 btw), and I immediately tried his suggestion of using
newsspool, only to find (which I should have known too, of course) that
I didn't get my beloved Lines: header...

This is of course a bit un-nice to "fix" in mail2news, as it requires
reading all of the input, however Neil also said (in email) that he is
planning to implement Message-ID-generated-from-body-contents-for-mail-
that-doesn't-have-one, which would mean having to read all of the input
anyway - said he, but now that I look again at the README, I note that
you mention using just the first ten lines - seems sensible to me, and
then I guess there's no hope for Lines: after all... (sob:-)

Anyway, I have now installed and am using this version, and I like it a
lot, in particular I think news2mail is now much better suited for
"local" gatewaying, which made me finally implement news->mail. But of
course, there's always those little things... - none of the stuff below
is of major importance, and I haven't made any diffs since I have
nothing to diff against, since it's mostly trivial things and you may
not like them anyway - but if you think they're reasonable and that
getting tested diffs would save you some time, I can certainly provide
such, although probably not before New Year. Here goes:

The C news getabsdate() stomps on the date string (NUL-terminates the
words), thus the "retry" and X-Unparseable-Date are no good, you need to
call getabsdate on a copy.

There are certainly systems where even gethostbyname(gethostname())
returns an unqualified name (such as the one I'm running this stuff
on:-), thus (just) doing this on local addresses is perhaps not a good
idea, in particular since inews on all systems I know of will provide a
full domain name for From: if it just has a username. Alternatives might
be to make this stuff conditional on DO_ADDRESS_CLEANUP, use WHOAMI if
defined (see below), or what I'd prefer (since it would handle non-local
but unqualified addresses too), have a simple DEFAULT_DOMAIN to be
applied if the address is unqualified after gethostname() and optional
gethostbyname(), like most newsreaders do when posting (might be
combined with the WHOAMI method).

I stumbled upon the lowercasing problem - Sun's stupid sendmail will
apparently lowercase the arguments when doing a queue run *regardless*
of whether the prog mailer has the "preserve case" flag set - however
the main problem this caused was not with the -o argument, but with -W
that I tried to pass to inews - it rejected the article altogether with
"bad option: -w":-(. Perhaps using inews -W isn't a terribly good idea,
however -F would have the same problem, and if you want to handle any
program/options, it seems that HackPeriods() is called for on *all*
argument words - although probably the arguments of -n, -x, -a, (-d?)
should be excluded for improved backwards compatibilty - what a mess...

It seems to me that the definition (or rather the use) of DO_8BIT_CHARS
is reversed?

That's all the "code" changes, the rest is mainly suggestions on
"improvements" in documentation/comments etc.

About -x etc: I whined a bit about this in news.software.b (in
<1991Dec15.111023.6227@eua.ericsson.se>) and how B and C news handles it
differently, with the only sys file setup working for B not working
for C and vice versa, and how the optimisation being discussed for C
that was in B while desirable would have the effect that -x didn't work
at all for this purpose, the bottom line being:

"One method that should work in all cases is to not bother with -x, use
the listname-out/gateway style, and force "gateway!..." into the Path:
of articles coming in from mail instead."

- perhaps this should be pointed out in the docs (I know that the method
is mentioned, but I think it might be a good idea to recommend it), and
the suggested value for FIXED_PATH modified slightly.

Of course, the above prompted a followup pointing out that there might
be a site 'gateway' out there, i.e. you should use something.your.domain
instead, and of course looking in the maps I find that this is actually
the case (Gateway Computing, NC, USA, aka gateway.com - I bet they don't
get a lot of news:-)! Thus recommending the use of a simple 'gateway' is
perhaps not a terribly good idea.

Another minor comment on gate.h:
/* If the "hostname" you use for news (such as in the pathline) is not
 * the same as `hostname`, then define WHOAMI to be your news name (e.g.,
 * the contents of the C News file /usr/lib/news/whoami). */
/* EUA/Per H: This is wrong, WHOAMI is only used by news2mail
   for the Received: and possibly the From: line,
   i.e. it should be the "mail name", not the "news name".       */

And one about README: I was going to point out that another reason for
using relaynews or something rather than inews is to avoid the "more
included than new" silliness, but apparently C inews doesn't do this,
which might be mentioned in the todo list. Also, it might be mentioned
somewhere here that e.g. relaynews cannot be used for the INEWS #define
(it wasn't obvious to me...:-).

Regards
--Per

PS I was going to bother you with this even if you hadn't posted on the
subject:-).


From kre@munnari.oz.au Sat Jan  4 07:04:24 1992
Received: from BBN.COM by pineapple.bbn.com id <AA20062@pineapple.bbn.com>; Sat, 4 Jan 92 07:04:19 -0500
Received: from munnari.OZ.AU by BBN.COM id aa07096; 4 Jan 92 7:04 EST
Received: from mundamutti.cs.mu.OZ.AU by munnari.oz.au with SMTP (5.64+1.3.1+0.50)
	id AA29525; Sat, 4 Jan 1992 23:03:58 +1100 (from kre)
To: rsalz
Subject: newsgate & inn
Date: Sat, 04 Jan 92 23:03:43 +0100
Message-Id: <15296.694526623@munnari.oz.au>
From: Robert Elz <kre@munnari.oz.au>
Status: RO

I just had a need to install newsgate (from c.s.u vol 24).
To make it work, I had to fix closefiles() in gag.y to
do the fprintf to "F" rather than "bbn" as for me "bbn" was
NULL...  But that one I expect someone else had told you
ages ago.

But I also needed to alter the sendmail aliases so they appear
as

	name: |"/.../ ..."

That is, with the '|' outside the quotes.  If its inside its just
part of trhe token, which works OK in normal sendmail.cf setups,
but on munnari I have to recognise all local "user" parts, and
ship the others off elsewhere .. that my sendmail.cf does by
looking up the "user" part in a database, if its found, the mail
is delivered (after aliasing, .forwarding, ...) otherwise an
"@host" is added, and the message is forwarded elsewhere.

Naturally that doesn't work well with mail to files or pipes,
as forwarding it elsewhere just generates the "you are not
allowed to mail directly to a program" diagnostic from the
other system, and the "|process..." certainly isn't going to
exist in the user database.

So, I need to handle the |... and /... parts separately, and
that means having the '/' and '|' be separate tokens for the
sendmail.cf - if they're inside the quotes they're not, if
they're outside the quotes they are.

I know this breaks the "can't quote anything but the whole local
part" rule, but sendmail doesn't care, and these things are never
seen anywhere else.

Next, I really wanted a mailing list that could cross post into
two newsgroups (yes, there's really a reason, its not really
a mailing list, just a method for one person to post).  That
worked OK by just setting

	inews flags = "-n grp1,grp2"

which caused gag to generate -n grpname -n grp1,grp2
and inews just uses the last "-n" arg it sees, so the one
coded in didn't hurt, but this seemed a luttle ineligant,
so I added ',' to the set of characters that represent tID
in lex.l so the ',' can appear in what gag thinks is a newsgroup
name - naturally this will fail gag's "does this newsgroup
exist" test, but -a to gag avoids that one...  (the test could
be coded to allow for multiple newsgroups, but I couldn't be
bothered).   Now I do

	gateway	grp1,grp2 listname

and get
	listname: |"/.../inews -n grp1,grp2 ..."

which is just fine.


Then I needed to be able to set the "Followup-To:" header
in the articles, so I usurped '-w' in inews to set that
header (both f and F were taken).   Defaulting to the first
group would do the wrong thing in the cross-posted case, and
its needed in any case when the group is moderated, and
followups should go to an unmoderated discussion group rather
than to the moderated group (which is exactly what should
happen in the case I care about today).

I gather from the README, etc, in newsgate that you don't
really want to bother much more with that one, but if you
see no objections, putting '-w' (or some other flag) in
inews in the next revision would make it easier for me to
install (yes, feeping creatureism...)

kre

From wolfgang%wsrcc.com@mescal.noc.vitalink.com Wed Apr  1 22:26:49 1992
Received: from relay2.UU.NET by rodan.UU.NET with SMTP 
	(5.61/UUNET-mail-drop) id AA27553; Wed, 1 Apr 92 22:26:49 -0500
Received: from world.std.com by relay2.UU.NET with SMTP 
	(5.61/UUNET-internet-primary) id AA03503; Wed, 1 Apr 92 22:26:57 -0500
Received: from BBN.COM by world.std.com (5.61+++/Spike-2.0)
	id AA17497; Wed, 1 Apr 92 22:26:34 -0500
Received: from mescal.NOC.Vitalink.COM by BBN.COM id aa07107; 1 Apr 92 22:25 EST
Received: by mescal.NOC.Vitalink.COM with UUCP id AA24153
  (5.65c/IDA-1.4.4 for rsalz@bbn.com); Wed, 1 Apr 1992 19:20:35 -0800
Received: by wsrcc.com id AA04316
  (5.65c/IDA-1.4.4-WSR-02/03/92 for rsalz@bbn.com); Wed, 1 Apr 1992 18:28:06 -0800
Date: Wed, 1 Apr 1992 18:28:06 -0800
From: wolfgang@wsrcc.com (Wolfgang S. Rupprecht)
Message-Id: <199204020228.AA04316@wsrcc.com>
To: rsalz@bbn.com
Subject: mail2news bug
Organization: W S Rupprecht Computer Consulting, Fremont CA
Status: R


Mail2news had problem with the following header.  Notice that the date
got mangled, which then caused C news to silently dump it.  Arrgg!
I'm beginning to think that important mail shouldn't be put into news.

    From scheidt!djs@vanbc.wimsey.bc.ca Wed Apr  1 17:19:00 1992
    Received: from vitanoc.UUCP by wsrcc.com with UUCP id AA03524
      (5.65c/IDA-1.4.4-WSR-02/03/92 for alison@wsrcc.com); Wed, 1 Apr 1992 17:18:58 -0800
    Received: from maccs.DCSS.McMaster.CA by mescal.NOC.Vitalink.COM with SMTP id AA20205
      (5.65c/IDA-1.4.4); Wed, 1 Apr 1992 14:52:52 -0800
    Received: by maccs.dcss.mcmaster.ca (/\==/\ Smail3.1.24.1 #24.2)
	    id <m0lWEBl-000gWRC@maccs.dcss.mcmaster.ca>; Wed, 1 Apr 92 17:53 EST
    Received: by vanbc.wimsey.bc.ca (/\==/\ Smail3.1.22.1 #22.7 rmail)
	    id <m0lWE9w-0002BTC@vanbc.wimsey.bc.ca>; Wed, 1 Apr 92 14:52 PST
    Received: by scheidt.wimsey.bc.ca (pcmail)     Apr 01 28:13:18 1992
    Date:     Apr 01 28:13:18 1992
    From: Daniel Scheidt <djs@scheidt.wimsey.bc.ca>
    Message-Id: <823@scheidt.wimsey.bc.ca>
    To: forth-edu-l@maccs.dcss.mcmaster.ca
    Subject: workshop announcement



    Approved: news@wsrcc.com
    Distribution: wsrcc-net
    Path: news-mail-gateway
    From: djs@scheidt.wimsey.bc.CA (Daniel Scheidt)
    Newsgroups: list.forth-edu
    Subject: workshop announcement
    Message-ID: <823@scheidt.wimsey.bc.ca>
    Date: 31 Dec 69 23:59:59 GMT
    Organization: News Gateway at WSRCC


---
Wolfgang Rupprecht    wolfgang@wsrcc.com (or) wsrcc!wolfgang
Snail Mail:           39469 Gallaudet Drive, Fremont, CA 94538-4511

From media!ka3ovk!jadpc!nosc!wsrcc.com!wolfgang Thu Apr  2 21:08:18 1992
Received: from relay1.UU.NET by rodan.UU.NET with SMTP 
	(5.61/UUNET-mail-drop) id AA07357; Thu, 2 Apr 92 21:08:17 -0500
Received: from uunet.uu.net (via LOCALHOST.UU.NET) by relay1.UU.NET with SMTP 
	(5.61/UUNET-internet-primary) id AA08416; Thu, 2 Apr 92 21:07:28 -0500
Received: from media.UUCP by uunet.uu.net with UUCP/RMAIL
	(queueing-rmail) id 204645.28778; Thu, 2 Apr 1992 20:46:45 EST
Received: by media.uucp (/\=-/\ Smail3.1.18.1 #18.3)
	id <m0lWYnY-0001IkC@media.uucp>; Thu, 2 Apr 92 15:54 EST
Received: by ka3ovk.uucp (/\=-/\ Smail3.1.18.1 #18.10)
	id <m0lWYJG-0001bkC@ka3ovk.uucp>; Thu, 2 Apr 92 15:22 EST
Received: by jadpc.cts.com (Smail3.1.21.1 with UUCP)
	id <m0lWWxv-0000zyC@jadpc.cts.com>; Thu, 2 Apr 92 10:56 PST
Received: from mescal.NOC.Vitalink.COM by trout.nosc.mil (5.59/1.27)
	id AA07154; Thu, 2 Apr 92 10:44:16 PST
Received: by mescal.NOC.Vitalink.COM with UUCP id AA07702
  (5.65c/IDA-1.4.4 for jadpc.cts.com!ka3ovk!media!uunet!rsalz); Thu, 2 Apr 1992 10:39:26 -0800
Received: by wsrcc.com id AA14900
  (5.65c/IDA-1.4.4-WSR-02/03/92 for media!uunet!rsalz@ka3ovk.UUCP (Rich Salz)); Thu, 2 Apr 1992 10:21:50 -0800
Date: Thu, 2 Apr 1992 10:21:50 -0800
From: wolfgang@wsrcc.com (Wolfgang S. Rupprecht)
Message-Id: <199204021821.AA14900@wsrcc.com>
To: ka3ovk!rsalz (Rich Salz)
In-Reply-To: <9204021252.AA29408@rodan.UU.NET>
Subject: Re:  mail2news bug
Organization: W S Rupprecht Computer Consulting, Fremont CA
Status: R


> I'm thinking about a rewrite that will do things like
> (if (!pasredate()) printf("x-original-date: %s")

Could I also convince you to just put a current date on the news if
none exists?  Right now it just gets silently dropped by C news as
"too old."

-wolfgang
---
Wolfgang Rupprecht    wolfgang@wsrcc.com (or) wsrcc!wolfgang
Snail Mail:           39469 Gallaudet Drive, Fremont, CA 94538-4511


From rsalz@osf.org Wed Jul 15 09:56:03 1992
Received: from relay2.UU.NET by rodan.UU.NET with SMTP 
	(5.61/UUNET-mail-drop) id AA14782; Wed, 15 Jul 92 09:56:02 -0400
Received: from postman.osf.org by relay2.UU.NET with SMTP 
	(5.61/UUNET-internet-primary) id AA19309; Wed, 15 Jul 92 09:56:05 -0400
Received: from earth.osf.org by postman.osf.org (5.64+/OSF 1.0)
	id AA04014; Wed, 15 Jul 92 09:55:57 -0400
Received: by earth.osf.org (5.64/4.7) id AA29406; Wed, 15 Jul 92 09:55:57 -0400
Date: Wed, 15 Jul 92 09:55:57 -0400
From: rsalz@osf.org
Message-Id: <9207151355.AA29406@earth.osf.org>
To: rsalz@uunet.UU.NET
Subject: newsgate bug
Status: R

>From rsalz Wed Jul 15 09:23:10 1992
Received: by earth.osf.org (5.64/4.7) id AA29278; Wed, 15 Jul 92 09:23:07 -0400
Date: Wed, 15 Jul 92 09:23:07 -0400
From: rsalz
Message-Id: <9207151323.AA29278@earth.osf.org>
To: rsalz
Subject: newsgate bug
Status: R

>From brendan@cs.widener.edu Tue Jul 14 19:54:56 1992
Received: from betty.cs.widener.edu by cs.widener.edu with SMTP id AA27632
  (5.65c/Widener-4.2 for <rsalz@osf.org>); Tue, 14 Jul 1992 19:54:35 -0400
Received: by betty.cs.widener.edu id AA12007
  (5.65c/Widener-4.1 for rsalz@osf.org); Tue, 14 Jul 1992 19:54:33 -0400
Date: Tue, 14 Jul 1992 19:54:33 -0400
From: Brendan Kehoe <brendan@cs.widener.edu>
Message-Id: <199207142354.AA12007@betty.cs.widener.edu>
To: rsalz@osf.org
Subject: news2mail buffer limit
Reply-To: brendan@cs.widener.edu
Status: RO


FYI, the From: line below was chopped short by news2mail.  As a
result, the mail sent out by it bounced, since it didn't have the
trailing parenthesis.  (If you've still got newsgate on a burner.)

B

--
Path: widener!hela.iti.org!usc!sdd.hp.com!spool.mu.edu!
From: schaker@leo.unm.edu (SchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSchakerSch)
Newsgroups: alt.tv.simpsons
Subject: Re: The Blackboard
Message-ID: <a9_m!=b@lynx.unm.edu>
Date: 14 Jul 92 21:34:18 GMT
Article-I.D.: lynx.a9_m!=b
References: <12601@hardy.rl.ac.uk>
Organization: University of New Mexico
Lines: 19

In article <12601@hardy.rl.ac.uk> rjm@inf.rl.ac.uk writes:
>Does anyone have any other ones to add to the list ?
>Here's the list......
> Hamsters cannot fly.

Soon after the hilarious Quayle incident we got a rerun with:

Potato, not potatoe.

[I'm pretty sure that's verbatim.]


				stef

-- 
    _---_     Stefan Chakerian
   / o o \    schaker@triton.unm.edu, schaker@unmb.bitnet
  | \___/ |   
   \_____/    Don't anthropomorphize computers.  They don't like it.



From chinacat!chip@cs.utexas.edu Fri Feb 28 23:10:56 1992
Received: by pineapple.bbn.com id <AA16942@pineapple.bbn.com>; Fri, 28 Feb 92 23:10:47 -0500
Received: from cs.utexas.edu by BBN.COM id aa26367; 28 Feb 92 23:08 EST
Received: from chinacat by cs.utexas.edu (5.64/1.120) with UUCP
	id AA00069; Fri, 28 Feb 92 21:46:27 -0600
Received: by chinacat.unicom.com (3.1.25.1) 
	id <m0lKKfm-00011PC@chinacat.unicom.com>; Fri, 28 Feb 92 21:23 CST
Message-Id: <m0lKKfm-00011PC@chinacat.unicom.com>
From: chip@chinacat.unicom.com (Chip Rosenthal)
Subject: newsgate
To: rsalz
Date: Fri, 28 Feb 92 21:23:42 CST
X-Mailer: ELM [version 2.3 PL11]
Status: RO

I have at least the mail->news end working on SCO XENIX 2.3.4, smail3.1.25,
and C news.  Since smail can be installed as /usr/lib/sendmail, I just
enabled the SENDMAIL definition.

Two things were required to get it going.

First, my cpp barfs on stuff like:

	#undef FOOBAR "blufl baz"

in gate.h and I rewrote all of them to:

	/*#define FOOBAR "blufl baz"	/**/

Second, even though I don't have a gethostname() (well, I do, but I
didn't want to compile in the socket library just to get it), I didn't
want to compile in a fixed host name.  I added the following to misc.c:

	#if	defined(GETHOSTNAME_UNAME)
	#include <sys/uname.h>
	int gethostname(buf, buflen)
	char *buf;
	int buflen;
	{
		struct utsname u;
		if (uname(&u) != 0 || sizeof(u.sysname)+1 > buflen)
			return -1;
		strncpy(buf, u.sysname, sizeof(u.sysname));
		buf[u.sysname] = '\0';
		return 0;
	}
	#endif

	#if	defined(GETHOSTNAME_FILE)
	int gethostname(buf, buflen)
	char *buf;
	int buflen;
	{
		FILE *fp;
		char *s;
		if ((fp = fopen(GETHOSTNAME_FILE, "r")) == NULL)
			return -1;
		if (fgets(buf, buflen, fp) == NULL) {
			(void) fclose(fp);
			return -1;
		}
		if ((s = RDX(buf, '\n')) != NULL)
			*s = '\0';
		(void) fclose(fp);
		return 0;
	}
	#endif

and to gate.h:

    /* Do you have a gethostname() call?  If not define one of the following.
     * GETHOSTNAME_UNAME uses the SysV uname() system call.  GETHOSTNAME_FILE
     * reads the hostname from the first line of the specified file (e.g. the
     * cnews "whoami" file, the XENIX /etc/systemid file, etc.) */
    /*#define GETHOSTNAME_UNAME	/**/
    #define GETHOSTNAME_FILE "/usenet/lib/news/whoami"	/**/


-- 
Chip Rosenthal   <chip@chinacat.Unicom.COM> | My mind's got a mind of its own.
Unicom Systems Development     512-482-8260 |   -Butch Hancock

From kyle@wendy-fate.uu.net Thu Jul 16 17:48:17 1992
Received: from wendy-fate.UU.NET by rodan.UU.NET with SMTP 
	(5.61/UUNET-mail-drop) id AA18996; Thu, 16 Jul 92 17:48:16 -0400
From: kyle@wendy-fate.uu.net (Kyle Jones)
Received: by wendy-fate.UU.NET (5.61/UUNET-mail-leaf)
	id AA19741; Thu, 16 Jul 92 17:47:59 -0400
Date: Thu, 16 Jul 92 17:47:59 -0400
Message-Id: <9207162147.AA19741@wendy-fate.UU.NET>
To: rsalz
Subject: mail2news, some comments
Status: R

Here's the comment from the top of my mail2news program.  Looking
at the man page and the code (searching for undocumented command
flags), I don't see any way to get the NEWS_SYS_TAG feature
described below without recompiling.  I think this should be a
runtime option.

Also being able to specify the ERROR_ADDRESS, and have mail2news
put this in the Sender header would be good.

/*
 * Usage: mail2news NEWS_SYS_TAG LIST_NAME ERROR_ADDRESS NEWSGROUPS [DISTRIBUTION]
 *
 * The mail message to be sent will read from stdin.  Unwanted
 * headers are stripped away, and the headers
 *
 *   Newsgroups: NEWSGROUPS
 *   Sender: ERROR_ADDRESS
 *   Path: NEWS_SYS_TAG!LIST_NAME
 *
 * are added.  If a Path header already exists, NEWS_SYSTAG will be
 * prepended to it, instead of creating a new header.  A pipe will
 * be opened to the RNEWS_COMMAND defined below and the message will
 * be passed to it.
 *
 * If the optional DISTRIBUTION argument is provided, a
 * Distribution: header will be added listing the provided
 * distribution.
 *
 * NEWS_SYS_TAG should match the first field of the entry in the
 * news sys file that gateways news into the mailing list (if such
 * gatewaying is done.)  This is what the sys entry should look like
 *
 * NEWS_SYS_TAG:world,NEWSGROUPS:S:/usr/lib/sendmail some-mailing-list
 *
 * The presence of NEWS_SYS_TAG in the Path header of the article
 * that this program generates will keep the article from looping
 * back into the mailing list.
 *
 * LIST_NAME can be any arbitrary string as long as it would be
 * syntactically correct for it to appear in the Path header as
 * shown above.
 */

From twinsun!twinsun.com!eggert@cs.ucla.edu Mon Jan 20 16:09:45 1992
Received: from BBN.COM by pineapple.bbn.com id <AA18547@pineapple.bbn.com>; Mon, 20 Jan 92 16:09:40 -0500
Received: from Kona.CS.UCLA.EDU by BBN.COM id aa24479; 20 Jan 92 16:07 EST
Received: from twinsun.UUCP by kona.cs.ucla.edu
	(Sendmail 5.61b+YP/3.17) id AA17506;
	Mon, 20 Jan 92 13:07:34 -0800
Received: from farside.twinsun.com by twinsun.com (4.1/SMI-4.1)
	id AA11217; Mon, 20 Jan 92 12:43:01 PST
Received: by farside.twinsun.com (4.1/SMI-4.1)
	id AA05466; Mon, 20 Jan 92 12:42:58 PST
Date: Mon, 20 Jan 92 12:42:58 PST
From: Paul Eggert <eggert@twinsun.com>
Message-Id: <9201202042.AA05466@farside.twinsun.com>
To: rsalz
Cc: info-mcl-request@cambridge.apple.com
Subject: newsgate problem with comp.lang.lisp.mcl posting
Status: R

Now that I know about the newsgate problem, I see other examples where the
newsgate Message-ID munger leads to duplicate articles.  For example, I just
noticed a duplicate posting as follows:

<47*/S=ban/OU=ifi/O=unizh/PRMD=SWITCH/ADMD=ARCOM/C=CH/@MHS> original Message-ID
<47*_S=ban_OU=ifi_O=unizh_PRMD=SWITCH_ADMD=ARCOM_C=CH_@MHS> munged Message-ID

The original Message-ID appeared in comp.lang.lisp.mcl, the munged duplicate
in comp.lang.lisp.  This Message-ID is apparently generated by a standard way
of transforming X.400 message ids into RFC 822 Message-IDs, so we can expect
to see more and more of these slashified monstrosities in the future.

>From the problems I've seen so far, I would advocate that newsgate never alter
a Message-ID that conforms to RFC 822 and 1036.  Newsgate shouldn't generate
duplicate articles just because some old broken news systems can't handle
valid Message-IDs containing `/' and `"'.

Here are copies of the headers in question.

Original:
	Path: twinsun!ucla-cs!usc!apple!cambridge.apple.com!ban@ifi.unizh.ch
	From: ban@ifi.unizh.ch (Bela Ban)
	Newsgroups: comp.lang.lisp.mcl
	Subject: Connections.lisp & CMWrite
	Message-ID: <47*/S=ban/OU=ifi/O=unizh/PRMD=SWITCH/ADMD=ARCOM/C=CH/@MHS>
	Date: 20 Jan 92 13:07:40 GMT
	Sender: info-mcl-request@cambridge.apple.com
	Lines: 27
	Approved: comp.lang.lisp.mcl@Cambridge.Apple.C0M
	X400-Received: by mta chx400.switch.ch in /PRMD=switch/ADMD=arcom/C=CH/;
	               Relayed; Mon, 20 Jan 1992 14: 16:58 +0100
	X400-Received: by /PRMD=SWITCH/ADMD=ARCOM/C=CH/; Relayed;
	               Mon, 20 Jan 1992 14: 07:40 +0100
	X400-Originator: ban@ifi.unizh.ch
	X400-Mts-Identifier: [/PRMD=SWITCH/ADMD=ARCOM/C=CH/;920120140740]
	X400-Content-Type: P2-1984 (2)
	Content-Identifier: 47
	Conversion: Prohibited
	Original-To: info-mcl@cambridge.apple.com

Duplicate generated by newsgate:

	Path: twinsun!ucla-cs!elroy.jpl.nasa.gov!swrinde!cs.utexas.edu!qt.cs.utexas.edu!yale.edu!yale!mintaka.lcs.mit.edu!gateway
	From: ban@ifi.unizh.ch (Bela Ban)
	Newsgroups: comp.lang.lisp
	Subject: Connections.lisp & CMWrite
	Message-ID: <47*_S=ban_OU=ifi_O=unizh_PRMD=SWITCH_ADMD=ARCOM_C=CH_@MHS>
	Date: 20 Jan 92 13:07:40 GMT
	Sender: newsgate@lcs.mit.edu
	Organization: LCS news/mail gateway
	Lines: 27

From @surname.draper.com:seb1525@mvs.draper.com Fri Aug  7 12:02:56 1992
Received: from relay1.UU.NET by rodan.UU.NET with SMTP 
	(5.61/UUNET-mail-drop) id AA18275; Fri, 7 Aug 92 12:02:48 -0400
Received: from ns.draper.com by relay1.UU.NET with SMTP 
	(5.61/UUNET-internet-primary) id AA05432; Fri, 7 Aug 92 12:02:42 -0400
Message-Id: <9208071602.AA05432@relay1.UU.NET>
Received: from surname.draper.com by ns.draper.com id aa13557;
          7 Aug 92 12:02 EDT
Received: from mvs.draper.com by surname.draper.com id aa04706;
          7 Aug 92 12:02 EDT
Received: from MVS by MVS.draper.com (IBM MVS SMTP V2R1) with BSMTP id 0644;
   Fri, 07 Aug 92 12:02:25 EDT
Received: by MVS.DRAPER.COM with NETMAIL; Fri, 07 Aug 1992 12:01:27 EDT
Date: Fri, 07 Aug 1992 12:01:00 EDT
From: "Stephen E. Bacher" <seb1525@mvs.draper.com>
Subject: mail2news patch for mailing to multiple newsgroups
To: rsalz@uunet.UU.NET, skh4161@mvs.draper.com
Status: RO



You may remember this...

We have set up mail2news in conjunction with C News and MMDF to route
mail from certain local mailing lists to local newsgroups.  We tried
this originally with a shell script that pipes the input through
a nawk program that appends the newsgroup name to the message ID and
thence to mail2news itself.

The problem was that occasionally the body of the message got truncated
or was completely missing.

After having experimented with a number of things, I came to the
conclusion that this would never work reliably.  It also may be
due to some anomaly of {n}awk generating non-null lines the way
we did things.  In any case, I took a much simpler approach:
I patched the C source code - specifically, rfc822.c - to hack
the message-id.

So far (knock wood) this seems to be working just fine.

We applied the patch to the module headed thusly:

 /nfs/papaya/u2/rsalz/src/newsgate/RCS/rfc822.c,
 v 1.11 91/07/22 10:26:51 rsalz Exp $

Here's the patch (hope it survives the trip :-):

--------------- cut here ----------------------------------------------
#!/bin/sh
cp rfc822.c rfc822.c.orig
ed -s rfc822.c <<THIS_IS_IT
509,510c
    if (hp->ident[0]) {					/* seb@draper.com */
	HackMessageId(hp->ident,hp->nbuf,hacked_message_id);
	Fprintf(fp, "Message-ID: %s\n", hacked_message_id);
    }
.
499a
    char		hacked_message_id[SM_SIZE];	/* seb@draper.com */
.
489a
#define HACK_MAX	21

void
HackMessageId(the_id, the_newsgroup, buf)
    char		*the_id;
    char		*the_newsgroup;
    char		*buf;
{
    char                *p;
    char                *q;
    char                *r;

 /* Find end of newsgroup name. */
 for (r = the_newsgroup;   *r != '\0' && !isspace(*r); r++);

 /* Point to portion of newsgroup name to insert in message-id. */
 if (r - the_newsgroup > HACK_MAX) r -= HACK_MAX;
 else r = the_newsgroup;

 /* Everything up to the "<" gets copied over. */
 for (p = the_id, q = buf; *p != '\0' && *p != '<'; p++, q++) *q = *p;
 *q++ = '<';

 /* Newsgroup name (or last HACK_MAX chars thereof) gets copied in. */
 for (; *r != '\0' && !isspace(*r); r++, q++) *q = *r;
 *q++ = '%';

 /* The rest of the message id gets copied over. */
 for (p++; *p != '\0'; p++, q++) *q = *p;
 *q = '\0';

 return;
}

.
488a
/*
 * seb@draper.com 5 Aug 1992: Hack message ID by inserting the newsgroup
 * name into it, to prevent mailings to multiple newsgroups from losing
 * the second and subsequent postings due to duplicate message ID.
 * This creates separate copies of the identical message, but it's still
 * better than having the article not show up at all in all the requested
 * groups.   So the message ID is hacked as follows:
 *
 * Old message ID:  <123456789@host.site.suf>
 * New message ID:  <newsgroup.name%123456789@host.site.suf>
 *
 * Note: If the newsgroup name is longer than HACK_MAX, only the last
 * HACK_MAX characters are hacked in.
 */
.
4a
 *
 * Modified by Steve Bacher <seb@draper.com> 5 August 1992 to implement
 * Message-Id hack for mailings to more than one newsgroup.
 *
.
w
q
THIS_IS_IT

--------------- cut here ----------------------------------------------

We will probably post this to USENET, with your permission.  Or perhaps
you would prefer to take the responsibility. :-)

Thanks for your help at IAUG.

                         -Steve Bacher
                        - seb@draper.com