NetNews Usenet Archive 1992 #27/NN_1992_27.iso/spool/comp/mail/uucp/2102

Internet Message Format | 1992-11-20 | 4.1 KB

Path: sparky!uunet!cs.utexas.edu!torn!nott!cunews!revcan!ecicrl!clewis
From: [e-mail] (Chris Lewis)
Newsgroups: comp.mail.uucp
Subject: Re: Any easy way to filter this type of UUCP traffic?
Message-ID: <[e-mail]>
Date: 21 Nov 92 03:07:34 GMT
References: <[e-mail]> <[e-mail]> <[e-mail]>
Organization: Elegant Communications Inc., Ottawa, Canada
Lines: 84

In article <[e-mail]> [e-mail] writes:
>In article <[e-mail]> [e-mail] (Chris Lewis) wrote:
>>In article <[e-mail]> [e-mail] writes:

>>>perception of why the maps exist; to provide an accurate model
>>>for UUCP network connectivity or to intentionally provide a
>>>distorted view of the UUCP network.

>>Not all links are bidirectional

>For replies, bounce messages etc. to work properly *for mail
>delivery via UUCP/rmail*, bidirectional links are required.

Actual bidirectional links are only required for bouncing when
stupid mailers are in the way.  Even smail 2.3 can cope with
*real* unidirectional links (uux says "no link").  Whether
pathalias knows it's a route is irrelevant, because the bounce
route is (usually, in the case of pure UUCP/dumb rmail) explicitly
requested by the return address uux argument.

>If the link isn't bidirectional, it is therefore unsuited for
>mail delivery via UUCP and should not be in the maps at all.

You're getting confused between the description of *publically* useable
links (the maps) and what is physically implemented (by UUCP).

Pathalias advertises public routes.  Not all routes.

If you place a link in the maps, you're explicitly offering it
to the rest of the world to use (within reason.  MBAS usage is
still frowned upon).

If I set my maps so that:

    - my feed marks me terminal
    - I don't list the backlink

I am explicitly indicating to the rest of the world that I can
be reached by that feed site (forward).  By not listing the backlink,
I'm omitting to tell the rest of the world that they can use me
to get to the feed site.  This doesn't say one way or the other
that it physically exists, but, with UUCP, it usually does.

In my local maps, I show the link to the feed site so that I
can use it.  No problem.

Other people can use it to send mail to me.  No problem.

If the mail to me bounces (they spelled my name wrong), the
bounce will *still* work because the link physically exists,
regardless of whether my machine reroutes the bounce or not.
In the former case, my local map override has the backlink explicitly
mentioned so it will often go back out the same way it came (but
not necessarily), in the latter case, the route was explicitly mentioned
and it will work whether pathalias thinks there's a route or not.

If someone has to use this backlink to get to the feed site,
it isn't *my* problem.  I didn't publically advertise the link.
*I don't want them to use it*.  100*DEADing isn't going to help
at all.  Just hide the routings with poor connectivity.

Pathalias -vv isn't going to show my site.  It will show the feed
site.  But it will *not* be a problem with my site and not necessarily
with the feed site either.

The sites where the implicit backlinks exist are usually not the problem.
The problem is with the sites that don't have proper connectivity and
have to force thru implicit backlinks.  pathalias -vv saying that a route
includes a backlink is a symptom of this problem, but the implicit
backlink itself *isn't* the problem, and by its very existance (or
non-existance depending on your POV ;-) you can detect possibly
bogus routing.  But the backlink itself isn't necessarily the problem.

>>Except that now it'll take two people to delete it instead of one.

>Grasping at straws, Chris?

On the contrary Bruce.  You omit most of my points, quote me out
of context, zero in on a single minor point, and then you accuse
*me* of grasping at straws?
--
Chris Lewis; [e-mail]; Phone: Canada 613 832-0541
Psroff 3.0 info: [e-mail]
Ferret list: [e-mail]