[net.news.sa] sendsys msgs and return paths

rees@apollo.uucp (Jim Rees) (10/13/86)

    Netnews control messages are propagated exactly like all other news
    articles, and when your site receives a 'sendsys' control message, the
    news software uses the string in the Path: field of the message as a
    return path to mail your sys file back to the site which made the request.  

Only if your software is stupid.  Sending mail along the news path is
anti-social at best, and wrong at worst.  The path only reflects the path
the news took, and does not necessarily reflect an optimal, or even a
valid, mail path.  Semi-stupid mailers (like the one we run) will look
up a path in a pathalias data base given the site name of the target.
Smart mailers will use Honeyman's pathparse to disambiguate the target
and look up an optimal (modulo GIGO) path to it.

dyer@spdcc.UUCP (Steve Dyer) (10/15/86)

>>Netnews control messages are propagated exactly like all other news
>>articles, and when your site receives a 'sendsys' control message, the
>>news software uses the string in the Path: field of the message as a
>>return path to mail your sys file back to the site which made the request.  
>
>Only if your software is stupid.  Sending mail along the news path is
>anti-social at best, and wrong at worst.  The path only reflects the path
>the news took, and does not necessarily reflect an optimal, or even a
>valid, mail path.  Semi-stupid mailers (like the one we run) will look
>up a path in a pathalias data base given the site name of the target.
>Smart mailers will use Honeyman's pathparse to disambiguate the target
>and look up an optimal (modulo GIGO) path to it.

Apparently the person bellyaching about the return paths isn't running a
'smart mailer' with INTERNET defined, which is the important point here.
Without INTERNET defined in the news software, running a vanilla UNIX
distribution (generic Sys V, Ultrix 1.2, etc.) without any mail hacks,
the behavior I describe above is exactly what you're going to get, stupid
or not.  I find it disingenuous at best to pontificate about 'anti-social
behavior' on the part of the news/mail interface when smart mailers are
still in the minority and most sites behave just as is described here.

I believe News 2.11 allows one to define INTERNET and have replies sent
via the backbone provided one has listed well-known paths to backbone sites
in a news data file.  (This is from a 1 minute perusal of the 2.11 sources
so I might be a bit off.)
-- 
Steve Dyer
dyer@harvard.HARVARD.EDU
{linus,wanginst,bbnccv,harvard,ima,ihnp4}!spdcc!dyer

mark@cbosgd.ATT.COM (Mark Horton) (10/15/86)

In article <30ac6214.1de6@apollo.uucp> rees@apollo.uucp (Jim Rees) writes:
>Only if your software is stupid.  Sending mail along the news path is
>anti-social at best, and wrong at worst.  The path only reflects the path
>the news took, and does not necessarily reflect an optimal, or even a
>valid, mail path.  Semi-stupid mailers (like the one we run) will look
>up a path in a pathalias data base given the site name of the target.
>Smart mailers will use Honeyman's pathparse to disambiguate the target
>and look up an optimal (modulo GIGO) path to it.

Actually, on a system with a smart mailer (e.g. netnews has INTERNET
defined) the reply will go back to the From line, not the Path line.
Assuming that the From line is valid, it doesn't matter what path
the message took.  If you run smart mail software like smail or uumail
on your system, this is easy.

I can't see how pathparse would be of any help in this situation.  The
major practical application of pathparse is figuring out what's going
on when there is mixed %@ and ! syntax in an address.  The Path line
always generates a left-to-right path.  The From line doesn't get munged
when netnews is transmitted (at least, it isn't supposed to, and I don't
know of any systems that do) and it's generally in pure @ syntax.  Since
most sites are currently only registered in UUCP, it does help to have
a mailer that understands user@host.UUCP, which mailers such as smail
and uumail do.

james@reality1.UUCP (james) (10/16/86)

In article <30ac6214.1de6@apollo.uucp>, rees@apollo.uucp (Jim Rees) writes:
> Only if your software is stupid.  Sending mail along the news path is
> anti-social at best, and wrong at worst.  The path only reflects the path
> the news took, and does not necessarily reflect an optimal, or even a
> valid, mail path.  Semi-stupid mailers (like the one we run) will look
> up a path in a pathalias data base given the site name of the target.
> Smart mailers will use Honeyman's pathparse to disambiguate the target
> and look up an optimal (modulo GIGO) path to it.

Unfortunately, sending mail along a news path is the highest probability
of success route.  As you point out, "optimal" != "valid" in paths.
Unfortunately, the real world being what it is, "optimal" (as defined by
mod.map postings) does not imply valid.  It doesn't imply optimal.

The problem is how reliable you consider the mod.map database to be.  Since
it is only as reliable as the individual site administrators, I view the data
with a good deal of mistrust.  The Path: data is known to recently have worked
in the opposite direction, and since I assume unidirectional or news-only
links are rare (the rarity of unidirectional links might be a bad assumption)
I would consider the Path: data reliable.  "A bird in the hand is worth two
in the bush..."
-- 
James R. Van Artsdalen    ...!ut-ngp!utastro!osi3b2!james    "Live Free or Die"