Path: utzoo!attcan!uunet!zaphod.mps.ohio-state.edu!usc!elroy.jpl.nasa.gov!zardoz.cpd.com!dhw68k!david
From: david@dhw68k.cts.com (David H. Wolfskill)
Newsgroups: comp.mail.uucp
Subject: Re: What should a new UUUCP do?
Summary: map entries show *outgoing* -- not *incoming* links
Message-ID: <1990Dec10.155237.13603@dhw68k.cts.com>
Date: 10 Dec 90 15:52:37 GMT
References: <1CE00001.ab22ep@tbomb.ice.com>
Organization: Wolfskill & Dowling residence; Anaheim, CA (USA)
Lines: 56

In an article, someone wrote:
>> why have
>> yourself listed in the maps as connecting to other systems with a
>> full map entry if you don't want mail mapped through you?

>This is required so that people know how to get mail *to* me as well
>as *through* me. If I don't list the hosts I talk to, you don't know
>how to get to me.

This, with respect to the use of mailers that make use of the output of
pathalias, is not the case.

Suppose we feed the following to pathalias as if we were at site foo1:

bar	foo1(DEMAND), foo2(DIRECT)
foo1	foo3(DEMAND), bar(DEMAND)

$ pathalias -l foo1 -c
0	foo1	%s
300	bar	bar!%s
500	foo2	bar!foo2!%s
300	foo3	foo3!%s

In the above, pathalias is showing that to get from foo1 to foo2 -- even
though there were no lines in the above that would rationally have been
in a map entry for "foo2."

On the other hand, it would be of some benefit to site "foo2" to have a
map entry:

* It would make it easier for an SA there to run pathalias -- though
  such a site might well use smail's "smart-host" specification.

* It would also make a "batter" claim to the valuable name :-) than the
  lack of a map entry would.

* It would allow some of us an opportunity to get some idea how
  serious the folks at foo2 are about UUCP mail.  :-)

>Now, there is a means of specifying that your host is a "leaf" node,
>allowing pathalias et.al. to determine not to route mail through you.

This is true, but only works well if *all* sites that show a link to you
use it -- if at least one site does not, pathalias tends to notice, and
route mail through the "terminal" site that way.  :-(  I find that in
practice, I use the "DEAD" specification (either "-d site" on the
command line or "dead {site}" in the pathalias input) in order to
achieve the desired effect of using the site in question as a "last
resort."

Cheers,
david
-- 
David H. Wolfskill
uucp: ...{spsd,zardoz,felix}!dhw68k!david	InterNet: david@dhw68k.cts.com
CompuServe: >internet:david@dhw68k.cts.com