On Jun 23, 13:11, Una Smith wrote:
} I don't know what ltr did, but I suspect ltr simply posted an FAQ that
} was identical to the previous version, perhaps too soon after the previous
} posting.  I have read everything I can find about LISTSERV administration,
} so now it is time to bow my head and ask humbly whether the checksum
} heuristic can be manipulated to relax the restriction on duplicate messages
} for SCIFAQ-L, and what the default parameters are now.
 
Adapted from "The Umpteen Commandments of News Transport Software" (copyright
Henry Spencer ;-), the original poster of a news article should
never get articles bounced back at them for any reason.  If for some
reason some transport mechanism (whether news, gateway, server whatever)
can't forward the message, the message should be dropped on the floor with
an error message sent to the administrator of the transport system.
 
The reason for this is quite straight forward - the error is usually
not the original poster's fault (presuming that the poster's software
generates valid RFCxxx articles), and the poster can't do anything
about the problem.  And the poster may get deluged by such error
messages.  Ie: the bogus "NDN" bounces people were getting from
busted BBS gateways.
 
The LISTSERV should notice that the article originated on news (ie:
presence of Newsgroups: lines or similar), and drop the article
if there's a problem with a diagnostic for the system administrator.
But not mail anything back to the originator.
 
Further, the LISTSERV (and indeed, any mailing list server) should arrange
things so that bounces of mail to *its* subscribers do not get mailed back
to the original poster (or submitter).  I think LISTSERV already does this
reasonably.
 
} ...  Please
} be patient, everyone:  the Usenet -> LISTSERV gateway for sci.answers
} is an important extension of the existing distribution system.  Many
} sites can not get Usenet feeds, for whatever reason, but do have Usenet
} software that can be used to organize and share LISTSERV mailing list
} subscriptions as though they were Usenet groups.  SCIFAQ-L is, as far
} as I know, the first gateway from Usenet to a mailing list, and the
} first mailing list gated to a *.answers newsgroup.
 
Don't panic.  I, at least, am not after you yet ;-)  But it
is important that the LISTSERV authors hear about this stuff.
 
This isn't the first USENET -> mailing list gateway by a long
shot.  There are quite a few.  And periodically one runs amuck and
does far nastier things than we've seen SCIFAQ-L do.
 
 
--
Chris Lewis; [log in to unmask]; Phone: Canada 613 832-0541
Psroff 3.0 info: [log in to unmask]
Ferret list: [log in to unmask]