On Mon, 10 Feb 1992 23:00:44 -0500 Tom Cervenka said:
 (text deleted...)
 
 Someone else may have volunteered this info already, but I haven't read
 all my mail yet, so here goes...
 
> Listserv seems to fail more thatn succeed in catching this problem.
 
 I own/co-own numerous lists on JHUVM and I haven't seen Listserv fail to
 recognize (and then distribute to the list) a mail delivery notice for a
 long time even though I also get several bounces a day.  In fact, I just
 posted a note to LSTSRV-L showing an example of a very non-standard
 message that Listserv *did* catch.  Why did I post it?  Because there
 are different levels of loop detection you can ask Listserv to apply to
 incoming mail (on a per-list basis).  These are selected with the list
 keyword "Loopcheck".  Unless you have a compelling reason to do
 otherwise, I feel it's silly to change it from the default value "Full".
 If your lists do not do "Full" checking that could be why Listserv is
 letting the rejections get out to the list.  If you are doing full
 checking and mail rejections are still getting out to the list, send me
 examples, I want to see them!  There are sites that use absurd rules to
 contruct mail rejections messages, but they are (thankfully) rare.  If
 you have a problem site on your list, I'd advise you to contact them and
 explain the problem.  And again, I'd be curious to see samples.
 
>                                            For instance, "joe@bit1" is in the
> list as such but mail from him is address from "[log in to unmask]" and when
> SEND=PRIVATE he gets rejected when posting. I think this happens when people
> subscribe via a tell command instead of sending mail.
 
 This can happen for a number of reasons, the "tell" vs "mail" scenario
 you mention is one of them.  *But* in that case it means that the people
 in charge of the Bitnet nodes in questions have failed to register an
 Internet equivalence for their node name.  In other words, Listserv checks
 for registered node name equivalents before rejecting such mail.  If
 people are sending mail from mis-configured Bitnet sites, flame the people
 that failed to register the node properly.  Listserv is not to blame for
 trusting the Bitnet database entries.  I've also seen cases where people
 send mail from 'clustered' systems that produce different node names
 depending on which workstation/system you actually are logged into.
 Again, I feel such systems somewhat at fault for having an odd, cavalier,
 attitude about mail service.  But it's not quite as clearcut as the Bitnet
 case.  In those cases you have to resort to the old trick of signing
 people up with *all* possible node names and "NOMAIL"ing all but one.
 It's truely gross, but if you want to send "Send=Private" I can't think
 of any other way to do it, short of telling people to send from the same
 workstation everytime.
 
> =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
> Tom Cervenka - Univ of Il at Chicago, Info Cntr, Phone 6-7739
> Internet: ctct100icvmc.aiss.uiuc.edu
 
 -jj
 
 PS - I realize that dealing with Bitnic's UPDATE server is torturous,
 (I updatted several JHU node entries last month, to add Internet name
 equates amoung other things), but it's a responsibility that people
 accept when they join the net.