SUMMARY: sendmail and aliases/mailing list question

From: Michael H. Morse (mmorse@z.nsf.gov)
Date: Fri Oct 04 1991 - 08:37:24 CDT


My original question:

> I would like to create a mailing list that only I can use. A system
> alias is preferable so that the recipients won't see a long list of
> addresses in the header. But I don't want other people to be able to
> use my list as a "mail-reflector" (either intentionally or
> accidently). I'd prefer *not* to use the "list: ;" construct because
> I'm not sure it's supported by sendmail (at least the configuration I
> currently use), and I'm afraid it may break primitive mailers that
> rarely see it.

> What are the options? Is there a common way of doing this?

Thanks for the many many quick responses.
The answer was surprisingly simple. There were two basic options:
  1. Use blind carbon copies.
  2. Use a feature of sendmail that lets you put anything you
     want in the header.

For the bcc: route, Andrew Partan put it most succinctly:

        Set up two 'lists':
                list:you
                list-real:real,list,of,folks,on,the,list
        and then send the messages:
                To: list
                Bcc: list-real:
        
        Works just fine.

I would caution people to include the "To:" although it's not, strictly
speaking, necessary. The problem is that if you don't, the message header
will have no addressing info in it, and although your message will be delivered
just fine, some mailers may add confusing messages such as
"Apparently-to:".

The second option is what I'm going to use. Although the results of
both options are virtually identical, this one seems more direct. You
simply give sendmail the header you want. So to duplicate Andrew's
example, you would:

         %/usr/lib/sendmail list-real
         To: list
         From: you
         ...

Note that both of these options only provide security through obscurity.
In other words, if you use a system alias for list-real, then others can
reflect off it if they find out the name. This was fine for my use, but
several people suggested the same general tactics as above, but using
personal aliases contained in .mailrc, which would truly safe-guard your
list.

Many others sent me good tips on aliases and other sendmail features, such
as where errors go to. I've enclosed one of these, since it might be
useful to others:

>From Bob Cunningham:

     Try this setup (example is for a mailing list called "mumble", with me
     as the administrtor) in your /etc/aliases:
     
     mumble-request: bob@kahala.soest.hawaii.edu
     owner-mumble:bob@kahala.soest.hawaii.edu
     mumble: :include:/usr/export/home/kahala/bob/maillists/mumble
     
     (where the actual addressees, one per line, each with a trailing ','
     except the very last, are in that file .../maillists/works.)
     
                                 [ I think this is a typo ^ ]

     For the above, all the messages will appear to come from (and all
     errors will be directed to) you personally. If, instead, you'd rather
     that the messages appear to come from "mumble-request", and errors
     likewise to to that alias, then use something like this instead:
     
     mumble-request: bob@kahala
     mumble: "|/usr/lib/sendmail -oi -fmumble-request@hawaii.edu mumble-out"
     mumble-out: :include: /home/kahala/bob/maillists/mumble

No one answered my hidden question, which is how long does sendmail need to
have the alias file. In other words, if you use the construct:
      list: :include: filename
can you create a list in filename, invoke /usr/lib/sendmail to send a
message, and then immediately delete the list? Or to put it another way, by
the time /usr/lib/sendmail exits has it scarfed up all the information from
filename that it will ever need? I am thinking of creating a database of
users, each of whom wants different types of messages. When I have a
message ready, I determine from the database who gets it, build the list,
then invoke sendmail. Will this work, or do I have to leave the list intact
until the message is completely delivered to all recipients?

Thanks to all that responded:

      "Brad L. Knowles" <blknowle@aragorn.jdssc.dca.mil>
      Ken Coley <force-c!doc!kcc@uunet.uu.net>
      "Jose A. Fernandez" <jaf@inference.com>
      John Mc Grath <ext_adm!jmcgrath@uunet.uu.net>
      Scott Brim <swb@mitchell.cit.cornell.edu>
      Ron Vasey <vasey@mcc.com>
      "John A. Shriver" <jas@proteon.com>
      Neil W Rickert <rickert@cs.niu.edu>
      long-morrow@cs.yale.edu
      Henry Holtzman <holtzman@media-lab.media.mit.edu>
      "Henry N. Holtzman" <holtzman@media-lab.media.mit.edu>
      Bob Cunningham <bob@kahala.soest.hawaii.edu>
      Pat Stephenson <patrick@cs.cornell.edu>
      Andrew Partan <asp@uunet.uu.net>
      "Steve D. Miller" <steve@umiacs.umd.edu>
      Lisa Hamet <hamet@chrpuser.gsfc.nasa.gov>

--Mike



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:18 CDT