argv@turnpike.Eng.Sun.COM (Dan Heller) (06/07/90)
In article Craig_Everhart@TRANSARC.COM writes: > This is how you allow those who receive a message (call it A) to send > their replies to all the folks who received message A. Do you expect > them to have exactly the same set of Mush aliases that you do? Sometimes people send mail to an entire list of people with the intent that the recipients should *not* be able to reply to everyone, just the original author. This is accomplished by having the user's alias name on the To: line and have that passed to the MTA. In mush, this is accompished by setting the variable, $no_expand Are you saying that this is illegal? I'm sure the RFC's provide some method for accomplishing the same task if this isn't legal. -- dan ---------------------------------------------------- O'Reilly && Associates argv@sun.com / argv@ora.com Opinions expressed reflect those of the author only.
Craig_Everhart@TRANSARC.COM (06/07/90)
> Excerpts from netnews.comp.mail.misc: 7-Jun-90 user's aliases on the To: > l.. Dan Heller@turnpike.Eng. (884) > Sometimes people send mail to an entire list of people with the > intent that the recipients should *not* be able to reply to everyone, > just the original author. This is accomplished by having the user's > alias name on the To: line and have that passed to the MTA. The only mechanism (that I know of) that RFC822 provides to allow somebody to ``hide'' a list of destination addresses is to rewrite the list as an empty group name. Thus, you might have some destination list To: foobar-list that could be rewritten before transmission as: To: Foobar-List:; instead of, say, To: Foobar-List: name1@host1, name2@host2, name3@host3; or whatever. In AMS's delivery system, the obvious way to keep people from seeing the list of destination addressees, if you really want to do that, is to mail through a distribution list and then delete or rename the distribution list. Thus, once mail has gone out to: To: +dist+/afs/andrew.cmu.edu/usr13/cfe/party.dl@andrew.cmu.edu all I have to do is delete the ~/party.dl file before any of the recipients gets to see it. Yes, all these hacks allow To: and CC: fields to be unreplyable in some special circumstances, but that's not the usual way to operate. Furthermore, in these contexts, message recipients have some clue that the To: list shouldn't be usable. I don't know if Mush's ``no_expand'' variable leaves such traces. My own opinions about mail gateways that make no effort to make To:/CC: fields usable are rather different. Craig
rsalz@bbn.com (Rich Salz) (06/27/90)
In <10128@ogicse.ogc.edu> schaefer@ogicse.ogc.edu (Barton E. Schaefer) writes: |Joe User (joe@joes.domain.org) has in his .mailrc (or .mushrc) file: | alias everybody fred mary tom peter@far.away.edu \ | henry@farther.off.com uunet!nowhere!jane |If he sets $no_expand in mush and gives the command | mail everybody |mush will send out a message with the headers (among others) | From: joe@joes.domain.org | To: everybody It is totally wrong to send out mail with an invalid address. If you must provide a "no_expand" sort of facility, leave OFF the to address and treat everyone as Bcc recipients. /rich $alz -- Please send comp.sources.unix-related mail to rsalz@uunet.uu.net. Use a domain-based address or give alternate paths, or you may lose out.
rad@puffin.uss.tek.com (Richard Doty) (06/28/90)
I think mailing lists are quite interesting, and after reading through RFC1123 I'd like to make my site do the "right thing". Trouble is, I'm a little unsure what that is. I grepped through the Berkeley cf files that came with 5.64, and also through the gatech collection, and could find no place where anything _local_ was done with a list. My hunch is that one uses a special local mailer definition to provide the unique list semantics. Could someone please post (or e-mail) a few details on how this is actually accomplished? Thanks, Richard Doty rad@puffin.uss.tek.com
rad@puffin.USS.TEK.COM (Richard Doty) (06/28/90)
I guess a clarification is in order. I do know what the usual mailing list is, where you put an alias in /usr/lib/aliases and forward the letter to the members of the alias. What I'd like to know is how people get mail _groups_ to work, i.e. To: foo: user1 user2@some.dom user3 ; The configs I have seen just forward these things along, none actually try to process them. Thanks, Richard.
argv@turnpike.Eng.Sun.COM (Dan Heller) (06/29/90)
In article <2705@litchi.bbn.com> rsalz@bbn.com (Rich Salz) writes: > |If he sets $no_expand in mush and gives the command > | mail everybody > |mush will send out a message with the headers (among others) > | From: joe@joes.domain.org > | To: everybody > > It is totally wrong to send out mail with an invalid address. You're reacting out of context and are giving readers a skewed view based on your statement. Bart said the above, but noted that we have acknowledged that it was wrong and are considering better options. The problem is not that simple to solve since many sites don't recognize or do the same thing with any arbitrary header. > If you must provide a "no_expand" sort of facility, leave OFF the to > address and treat everyone as Bcc recipients. That's silly -- you must have a To: line or most MTAs such as sendmail will add one for you in the form of Apparently-To: and that's not going to contain very useful information either. The bottom line is, Mush's solution is "adequate" in that it doesn't provide any more or less than what other mailing lists do. However, it's not good enough in that it doesn't conform to the RFC. -- dan ---------------------------------------------------- O'Reilly && Associates argv@sun.com / argv@ora.com Opinions expressed reflect those of the author only.
Craig_Everhart@TRANSARC.COM (06/29/90)
> Excerpts from netnews.comp.mail.misc: 28-Jun-90 Re: user's aliases on > the T.. Richard Doty@puffin.USS. (396) > What I'd like to know is how > people get mail _groups_ to work, i.e. > To: foo: user1 user2@some.dom user3 ; > The configs I have seen just forward these things along, none > actually try to process them. First, let's point out that the proper syntax (1) requires commas between addressees and (2) requires host names on each addressee. Thus, the example should be: To: foo: user1@dom.one, user2@some.dom, user3@dom.xxx ; Having said that, what do you mean by ``get mail groups to work''? What do you believe is supposed to happen? What I believe should happen is that mail to the above list would be delivered to the three addressees ``user1@dom.one'', ``user2@some.dom'', and ``user3@dom.xxx''. And, yes, the addresses are supposed to stay intact in the actual message header text, though I suppose they wouldn't have to if you wanted your MUA to remove them. Craig
schaefer@ogicse.ogc.edu (Barton E. Schaefer) (06/30/90)
In article <2705@litchi.bbn.com> rsalz@bbn.com (Rich Salz) writes: } In <10128@ogicse.ogc.edu> schaefer@ogicse.ogc.edu (Barton E. Schaefer) writes: } |Joe User (joe@joes.domain.org) has in his .mailrc (or .mushrc) file: } | alias everybody fred mary tom peter@far.away.edu \ } | henry@farther.off.com uunet!nowhere!jane } |If he sets $no_expand in mush and gives the command } | mail everybody } |mush will send out a message with the headers (among others) } | From: joe@joes.domain.org } | To: everybody } } If you must provide a "no_expand" sort of facility, leave OFF the to } address and treat everyone as Bcc recipients. Unfortunately, that doesn't achieve the desired result, because sendmail will proceed to add one Apparently-To: line for every person to whom the mail is being sent. Not only does this often result in several hundred lines of headers, it also advertises the names of everyone on the list, which is generally the very thing the sender is trying to prevent. -- Bart Schaefer schaefer@cse.ogi.edu
rad@puffin.USS.TEK.COM (Richard Doty) (06/30/90)
In article <8aWqJZv0BwwOES_4p_@transarc.com> Craig_Everhart@TRANSARC.COM writes: >First, let's point out that the proper syntax (1) requires commas >between addressees and (2) requires host names on each addressee. Thus, >the example should be: > > To: foo: user1@dom.one, user2@some.dom, user3@dom.xxx ; I agree with the commas, but the individual destinations are mailboxes, so I would think they could be local addresses, too. >Having said that, what do you mean by ``get mail groups to work''? What >do you believe is supposed to happen? What I believe should happen is >that mail to the above list would be delivered to the three addressees >``user1@dom.one'', ``user2@some.dom'', and ``user3@dom.xxx''. And, yes, >the addresses are supposed to stay intact in the actual message header >text, though I suppose they wouldn't have to if you wanted your MUA to >remove them. According to RFC1123, When a message is delivered or forwarded to each address of an expanded list form, the return address in the envelope ("MAIL FROM:") MUST be changed to be the address of a person who administers the list, but the message header MUST be left unchanged; in particular, the "From" field of the message is unaffected. That's what I expect to happen. The only reference to list syntax that I am familiar with in sendmail.cf is this from S3: R$+:$*;@$+ $@$1:$2;@$3 list syntax R$+:$*; $@$1:$2; list syntax So suppose my local mailer receives a message for bind-hacks:;@puffin.uss.tek.com I am still not clear how people handle list format, without a unique mailer definition (at the least). Bill Wisner says list format is used commonly, but I've never seen it so I guess I've been "haunting the wrong corners of the Internet all this time..." :-) Anyway, I'd still like to see a real code example. Richard.
peter@ficc.ferranti.com (Peter da Silva) (06/30/90)
In article <8aWqJZv0BwwOES_4p_@transarc.com> Craig_Everhart@TRANSARC.COM writes: > To: foo: user1@dom.one, user2@some.dom, user3@dom.xxx ; > Having said that, what do you mean by ``get mail groups to work''? I don't know about getting them to work, but I'd like to know what the point of this syntax is. How does it differ, functionally, from: To: user1@dom.one, user2@some.dom, user3@dom.xxx Near as I can figure, these two forms should be interpreted the same way by the MUA, the MTA, and gateways alike. -- Peter da Silva. `-_-' +1 713 274 5180. <peter@ficc.ferranti.com>
Craig_Everhart@TRANSARC.COM (07/03/90)
> Excerpts from netnews.comp.mail.misc: 29-Jun-90 Re: user's aliases on > the T.. Richard Doty@puffin.USS. (1887) > >. Thus, > >the example should be: > > > > To: foo: user1@dom.one, user2@some.dom, user3@dom.xxx ; > I agree with the commas, but the individual destinations are > mailboxes, so I would think they could be local addresses, too. Sure, they could be local, but don't you always append the local domain to each addressee? You certainly have to when you send such mail off your site. > Excerpts from netnews.comp.mail.misc: 29-Jun-90 Re: user's aliases on > the T.. Richard Doty@puffin.USS. (1887) > According to RFC1123, > When a message is delivered or forwarded to each address of an > expanded list form, the return address in the envelope ("MAIL > FROM:") MUST be changed to be the address of a person who > administers the list, but the message header MUST be left > unchanged; in particular, the "From" field of the message is > unaffected. Ah, but you're mixing apples and lightbulbs here. The address-list syntax that includes a group name (as in ``name: <list>;'') has nothing particularly to do with an ``expanded list form'' as described in RFC 1123. The former is a way of syntactically grouping a list of addresses in a message header, and has limited utility to some user agents in allowing aggregation of sets of addressees. The latter refers to a real live mail redistribution mechanism, for example something that turns mail to header-people@mc.lcs.mit.edu to a message directed at lots of speciflc mailboxes (more specifically, mail to info-andrew@andrew.cmu.edu into a message addressed to about 200 people). About the only direct connection between these two forms is that some implementors of redistributing mechanisms have considered rewriting message headers of mail addressed to the list to be in the empty-group format (``name:;''). Of course, most mail redistributing mechanisms have learned to keep their mitts off the actual header. But where this discussion started was not with mail redistribution mechanisms, but rather with mass mailings directed by a user agent (e.g. Mush). If a real mail redistribution mechanism were involved, there would be no question about what to put in the To: and CC: fields of such mail: those fields would simply bear the address of the mail redistributor (e.g. ``info-andrew@andrew.cmu.edu''). The issue at hand is what To: and CC: should contain if the wide distribution comes from a user agent expanding a personal alias that lists a lot of names. Apparently, Mush, under some $no_expand flag, leaves only the name of the user's alias in the header, in spite of the fact that such an address is unreplyable--mail to that address, e.g. from off-site, won't be redistributed. One of the many alternatives to such a scheme for what-to-put-in-the-To:-field is to use one of these empty groups, so that recipients will at least know that the exact membership of the To: field is hidden to them, and that they can't send messages directly to ``the set of folks that got this one.'' Does this help? It doesn't make sense to compare syntactically-named address groups (name: addr, addr, addr, addr ;) with mail redistribution agents. They're not meant to be the same thing. Craig
david@twg.com (David S. Herron) (07/06/90)
In article <10128@ogicse.ogc.edu> schaefer@ogicse.ogc.edu (Barton E. Schaefer) writes: >In article <7329@gollum.twg.com> david@twg.com (David S. Herron) writes: >} In article <136802@sun.Eng.Sun.COM> argv@turnpike.Eng.Sun.COM (Dan Heller) writes: >} >Sometimes people send mail to an entire list of people with the >} >intent that the recipients should *not* be able to reply to everyone, >} >just the original author. ... >} The RFCs (822 to be exact) specify a form >} list-name: ; >} to be used for the purpose you're talking about. >} Does mush generate this when $no_expand is set? ... >Joe User (joe@joes.domain.org) has in his .mailrc (or .mushrc) file: > > alias everybody fred mary tom peter@far.away.edu \ > henry@farther.off.com uunet!nowhere!jane > >If he sets $no_expand in mush and gives the command > > mail everybody > >mush will send out a message with the headers (among others) > > From: joe@joes.domain.org > To: everybody EEG erm.. that is indistinguishable from an un-domained address. I personally feel it is a very very bad thing to EVER generate mail with un-domained addresses in the header. I know that within some local net people generally have mail floating around that doesn't have domains attached to the Mail IDs ... and that's fine so long as the mail stays in the local net. But once it escapes, and it may well escape without passing through a gateway which cleans up and domainifies the header, then it will only cause confusion to people replaying to a header with all these addresses which look like they're local. For instance: the security mailing list at "zardoz" sends out a normal digest format with the headers inside the digest being in UUCP format relative to zardoz. I use mh for reading mail and use the burst command to burst digests. So now I have a bunch of messages in the zardoz-security-list folder with unusable headers. Even if the digest had passed through a gateway machine which cleans the headers it wouldn't have gone into the body and cleaned up the headers in the body. Ergo: Always generate mail with full domain names in the header or in some other way avoid strenuously making it look as if the names in the headers are local. >Yes, I agree this is imperfect, and we have considered changes. It >isn't clear what the translation into 822 syntax should be; do we >make it > > To: everybody:; Sure.. Some other places do this. CSNET, for instance, recently started using this format for sending out the CSNET Forum type stuff. RFC-822 describes groups as so: 6.2.6. MULTIPLE MAILBOXES ... A set of individuals may wish to receive mail as a single unit (i.e., a distribution list). The <group> construct permits specification of such a list. Recipient mailboxes are speci- fied within the bracketed part (":" - ";"). A copy of the transmitted message is to be sent to each mailbox listed. This standard does not permit recursive specification of groups within groups. And an example A.2.4. Committee activity, with one author George is a member of a committee. He wishes to have any replies to his message go to all committee members. From: George Jones <Jones@Host.Net> Sender: Jones@Host Reply-To: The Committee: Jones@Host.Net, Smith@Other.Org, Doe@Somewhere-Else; Elsewhere it says that the list need not be present. In other words.. this group:; format is meant for the exact situation which the "alias" in mush and/or ucbmail covers. I'd propose two commands and/or variables use_group_syntax list_group_members -- <- David Herron, an MMDF weenie, <david@twg.com> <- Formerly: David Herron -- NonResident E-Mail Hack <david@ms.uky.edu> <- <- Sign me up for one "I survived Jaka's Story" T-shirt!
tneff@bfmny0.BFM.COM (Tom Neff) (07/06/90)
In article <7489@gollum.twg.com> david@twg.com (David S. Herron) writes: ... >RFC-822 describes groups as so: ... > From: George Jones <Jones@Host.Net> > Sender: Jones@Host > Reply-To: The Committee: Jones@Host.Net, > Smith@Other.Org, > Doe@Somewhere-Else; > >Elsewhere it says that the list need not be present. > >In other words.. this group:; format is meant for the exact situation >which the "alias" in mush and/or ucbmail covers. But this appears useless for real net-wide mailing lists. First of all the To: line is not specified in the above example, which was the original point of this thread. Secondly, if you omit the list of addresses in the Reply-To: header above, as the standard appears to allow, then you don't have anything a recipient user can actually reply to! Group names are not FQDN's, and others' machines cannot be expected to have the right alias for "The Committee:". I have hundreds of users on my lists, including UUCP, Internet, BITNET, MILNET, EARNET, MCI Mail, CompuServe and elsewhere. This is what I do: From: Joe Contributor <joeblow@imelda.street.edu> Sender: LEPROSY-REQUEST@bfmny0.BFM.COM Errors-To: LEPROSY-REQUEST@bfmny0.BFM.COM Reply-To: LEPROSY@bfmny0.BFM.COM To: LEPROSY (Hansens Disease Interest Group) That (plus date and message-id) is the header. Then in the envelope (implemented as an 'rmail' command) I list all the individual addressees: rmail fred@schmoo.bitnet blah!blit!bloop alma@frat.quayle.edu ... My usual mail feed has a 2K limit on uux commands, so I split up the envelope list as needed and make multiple submissions. Now I don't know if this satisfies all the theoretical gurus, but it WORKS. -- A doubled signature is the devil's work. ** Tom Neff <tneff@bfmny0.BFM.COM> -- A doubled signature is the devil's work. ** Tom Neff <tneff@bfmny0.BFM.COM>
tneff@bfmny0.BFM.COM (Tom Neff) (07/06/90)
In article <19900706145229.8.7THSON@GLOWWORM.LispM.SLCS.SLB.COM> 7thSon@SLCS.SLB.COM (Chris Garrigues) writes: >[I showed my sample header:] >> From: Joe Contributor <joeblow@imelda.street.edu> >> Sender: LEPROSY-REQUEST@bfmny0.BFM.COM >> Errors-To: LEPROSY-REQUEST@bfmny0.BFM.COM >> Reply-To: LEPROSY@bfmny0.BFM.COM >> To: LEPROSY (Hansens Disease Interest Group) > >The right thing to do in this case is: > > From: Joe Contributor <joeblow@imelda.street.edu> > Sender: LEPROSY-REQUEST@bfmny0.BFM.COM > Errors-To: LEPROSY-REQUEST@bfmny0.BFM.COM > To: LEPROSY@bfmny0.BFM.COM (Hansens Disease Interest Group) > >Your version works, but it really it is one more line than necessary. I add the "Reply-To: list@address" line so that subscribers' mailers will tend to send R[eplies] back to the entire list by default, rather than only to the original contributor, which seems to be what happens if only the "From: Joe Contributor" line is present. >Besides, didn't the original sender provide a TO line that looked like >"LEPROSY@bfmny0.BFM.COM" anyway? Yes -- it routed the contribution to my forwarding agent in the first place. But once I've forwarded it out to the whole list, I don't want it looping BACK to me by accident, as seemed to happen every so often courtesy of someone's homicidal mailer. So I omit the FQDN from the To: line. It serves no purpose there anyway -- the origin fields are the ones that need to be replyable. -- Canada -- a few acres of snow. ^v^v^ Tom Neff -- Voltaire v^v^v tneff@bfmny0.BFM.COM
7thSon@SLCS.SLB.COM (Chris Garrigues) (07/06/90)
From: tneff@bfmny0.BFM.COM
Date: Thu, 5 Jul 90 19:31 CDT
I have hundreds of users on my lists, including UUCP, Internet, BITNET,
MILNET, EARNET, MCI Mail, CompuServe and elsewhere. This is what I do:
From: Joe Contributor <joeblow@imelda.street.edu>
Sender: LEPROSY-REQUEST@bfmny0.BFM.COM
Errors-To: LEPROSY-REQUEST@bfmny0.BFM.COM
Reply-To: LEPROSY@bfmny0.BFM.COM
To: LEPROSY (Hansens Disease Interest Group)
The right thing to do in this case is:
From: Joe Contributor <joeblow@imelda.street.edu>
Sender: LEPROSY-REQUEST@bfmny0.BFM.COM
Errors-To: LEPROSY-REQUEST@bfmny0.BFM.COM
To: LEPROSY@bfmny0.BFM.COM (Hansens Disease Interest Group)
The :; case is for lists which are user specified (usually in personal
configuration files of one sort or another) and CANNOT be replied to.
Your version works, but it really it is one more line than necessary.
Besides, didn't the original sender provide a TO line that looked like
"LEPROSY@bfmny0.BFM.COM" anyway?
Chris Garrigues,
7thSon@SLCS.SLB.COMCraig_Everhart@TRANSARC.COM (07/17/90)
> Excerpts from netnews.comp.mail.misc: 16-Jul-90 Re: user's aliases on > the T.. Tom Neff@bfmny0.BFM.COM (2587) > Hey, there are 1.0E+6 mail user agents out there. It's not my fault if > some particular program (to which source is presumably available) chooses > to let the user choose between "From:" and "To:" but not between "From:" > and "Reply-To:". Patch the program or switch. RFC 822 says that replies-to-sender go to Reply-to:, not From: (unless Reply-to: is absent). Look it up--section 4.4.4. > If Reply-To: already contains the list address in replyable form, why > does the To: field need to hold the identical contents. I've already argued that putting the list-submission address in Reply-to: is a silly idea. What you're doing is crowbarring the To: field to make it unusable, which is not playing the game. RFC 822's syntax for the To: field, as transmitted on the wire, doesn't admit the form that omits the ``@'' and domain name. Craig