LSTSRV-L Archives

LISTSERV Site Administrators' Forum

LSTSRV-L

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Topic: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Eric Thomas <ERIC@FRECP11>
Wed, 12 Nov 1986 15:22 SET
text/plain (67 lines)
  Here is what happened on the IRLEARN  loop problem... Niall had set his INFO-
VAX list to "Send= Editor"  (ie moderated list) with "Editor= [log in to unmask]
ARPA". When you set "Send= Editor" and "Editor= ed1,ed2,...", any of the people
listed in the "Editor="  keyword can mail directly to the  list while mail from
any other network address is transferred  to "ed1", the "main editor". In other
words, "ed1" is the list moderator while "ed2,ed3..." are "trusted" people from
which mail should always be accepted  without being first submitted to the list
moderator.
 
  With "Editor= INFO-VAX@...", everything  sent to INFO-VAX@IRLEARN was forwar-
ded to the master list on ARPAland, except  if it came from the master list it-
self. At least, that's how Niall thought it would have worked.
 
  Unfortunately mail from  ARPA lists does not have the  list-userid as "From:"
nor "Sender:".  It is  usually in the  "To:" field, although  I have  seen some
pieces of ARPA mail *without* a "To:". Since the "Editor" feature was NOT desi-
gned to  operate as  an automatic-forwarder  process for  hierarchically linked
lists, it  looks at the (Resent-)"From:"  and "Sender:" tags to  determine from
whom the piece of mail came. The "To:" field is assumed to be the list address,
ie INFO-VAX@IRLEARN  in our case. I  personally do not see  anything wrong with
that, nor do I see any reason  why it should be changed to operate differently.
 
  Another problem  was that the  process of resending a  file to the  editor is
done via a "SENDFILE"-like  process, ie as a file, not  through the MAILER. The
reason for  that is  that it is  not yet  possible to use  BSMTP to  resend the
mailfile 'as  is' to the  editor (if the destination  mailer uses BITNET  2 and
gets a BSMTP-sent  mail with a wrong  "To:" header, it will  not be distributed
properly) and I did not want to  add "Resent-" tags since there could have been
"Resent-" tags in the original piece of mail and these would need to be preser-
ved (otherwise the editor would not  know the origin of the mailfile). Granted,
I could have changed  the "To:" or "Resent-To:" field to  point to the editor's
mailbox, but then he  would not have known any more to which  list the mail was
aimed. I would have had to add a line of "Comment: This mailfile was originally
sent to list XXXXXX", which would then have appeared on the header of the mail-
file when  (and if)  the editor  decides to  resend the  mailfile to  the list.
 
  Anyway, the  process currently assumed  the editor  was a BITNET  person. The
idea behind this is  that if a BITNET list must be editored  by an ARPA-man, it
would be  much easier  to send to  the ARPA-list directly  and have  the editor
submit to the BITNET list with  "Send= Owner". The "Internal error" reported by
LISTSERV was due to the mail-routine which improperly considered the informati-
ve return  code "Note: your file  was sent through the  mailer because SENDFILE
could not  be used" as  an error because it  didn't expect ARPA  userids. There
had been no internal error at all, and the file-sending routine had generated a
mail envelope to send the "file" to  the ARPA recipient -- hence the subject of
File "xxx xxxx" being sent to you.
 
  That's what  I wanted  to say. I'd  like to add  that FRECP11-LISTSERV  has a
built-in routine that automatically detects rejection mail of (nearly) any kind
being sent  to it. The  BITNIC-type servers take  steps to have  rejection mail
sent to  another userid (and this  has unpleasant side effects  on the MAILBOOK
software package, but let's not start  a polemics about this again); however if
it receives a rejection mail from a  mailer which does not fully respect RFC822
it gets  into a mailing  loop (remember the one  on EARNTECH just  before CEARN
installed FRECP11-LISTSERV?).  Should my  software be  blamed for  the INFO-VAX
loop? I don't think so. If someone sets a list pointing to itself using BITNIC-
LISTSERV and generates a huge loop, I guess he would be held responsible for it
(don't laugh, this has already happened  with one of the FRECP11-LISTSERVs, but
hopefully FRECP11-LISTSERV does not enter a loop when a list points to itself).
 
Regards,
         Eric
 
PS: I am not  on the MAIL-L list. Please send replies directly  to me. To avoid
    further protestations:  I did not set  a "Reply-To:" tag pointing  to me in
    this  piece of  mail  since  the BITNIC  server  would  discard it  anyway.

ATOM RSS1 RSS2