🚀 go-pugleaf

In article <19920923080621SEB1525@MVS.draper.com>, SEB1525@MVS.draper.com (Steve
Bacher) says:
>I've already contacted Jim McIntosh and sent him some documentation
>comparing our old direct BITNET mailing list feed with our USENET feed
>for selected bit.listserv.* groups.  Some things have improved -
>possibly independently of any investigation - but the only specific
>I know of is the fixing of Alan Flavell's mailer, which Alan and I
>hashed out together.

Right.  The  circumstances  were  that the  sender's  mail  software  was
generating  an illegal  (by Usenet  standards) Message-ID.  The ListServ,
which could  care less about  Message-ID's, replicated the  Message-ID on
the post sent to the NetNews  server because NetNews has its subscription
set to FULLHDR. (NetNews depends on  Message-ID being present in the post
to discard  the post  when it comes  back from the  list, and  to control
duplicates in a situation where there may be two sites acting as a global
gateway.)

NetNews has now  received a post with an invalid  Message-ID. Rather than
discarding the post, which would result in 100% loss, or mucking with the
Message-ID, and thereby possibly contribution  to a mail loop, the server
accepts the post and sends a copy to nntp to be distributed to the world.
When nntp attempts  to distribute the post, however,  the receiving sites
detect the  invalid Message-ID and reject  the feed. It is  at that point
that the post is lost on its way from the mailing list to Usenet.

On one hand  the original sender has generated an  illegal Message-ID and
the solution is  for them to fix  their software. On the  other hand, the
post was accepted to the mailing list so the gateway should find some way
to get the post to Usenet. On the third hand... no, that won't work, will
it? :-)

If Linda  were to start  parsing Message-ID's and detecting  invalid ones
(using the  same rules as  nntp uses), and then  generating a new  one in
those cases  where the Message-ID  was invalid,  can anyone think  of any
dangers outside of a possible duplicate  post? Could a mail loop develop?
I don't think so, but let's have some other people also think it through.

If no one can  find any danger in doing this, then adding  a check on the
Message-ID  for mail  coming from  a  mailing list,  and correcting  that
Message-ID, would  result in less  (and possibly eliminate)  mailing list
posts not making their way to Usenet.
---
Jim McIntosh (jim@american.edu)
The American University
Washington DC 20016-8019 USA