RFC: Bionet <=> LISTSERV gateway

[Una Smith]

This is a semi-formal request for comments (RFC).  Below, I have outlined
a proposal for establishing gateways between existing "LISTSERV" mailing
lists and bionet newsgroups:


	1. Existing biology-related listserver mailing lists should be
	   allowed to establish gateways into bionet newsgroups, after
	   suitable discussion and approval in bionet and the list. 

	2. If there exists an appropriate bionet group, that group should
    	   discuss and vote on the proposed gateway, along with the members
	   of the mailing list.

	3. If no appropriate group exists, a discussion and vote should be
	   held in bionet, as is now done for "regular" new bionet groups.

	4. At the discretion of the owner and members of the listserv
	   mailing list, either

		4a. BIOSCI should take over administration of the list
		    (as is now done for the 40 bionet groups) or

		4b. a gateway should be opened (as is now done for the 200
		    bit.listserv groups), preferably at AUVM.american.edu.  

	5. Existing Usenet-style voting procedures used by the bionet 
	   community should be used to ensure that a proposed gateway meets
	   the requirements of the majority of people involved, including 
	   the administrators of both the list and the newsgroup.  The
	   mailing list should also conduct a survey and/or vote.

	6. Subject to the approval of all parties involved, non-listserv
	   mailing lists that wish to join the bionet groups shall either

		6a. become listserv lists (not necessarily running on an
		    IBM mainframe, but using identical protocols), or

		6b. turn over maintenance to BIOSCI.

	7. Only one gateway should exist.  All subscriptions for a particular
	   group should go to a single address.


Comments:

AUVM.american.edu is where most of the bit.listserv groups are gated, as
well as a handful of listserver mailing lists that successfully went
through and passed the voting process to become full Usenet newsgroups
(eg, GIS-L at UBVM.cc.buffalo.edu <=> comp.infosystems.gis).  AUVM has a
proven track record of maintaining these gateways in as seamless and
bounce-free a manner as have the BIOSCI administrators.  One thing that
the NETNEWS maintainers there suggest is that only one gateway be used,
and I think that's good advice, thus item 7 above.

Some background information about this proposal:  In 1987 a friend of
mine wrote NETNEWS, a Usenet server (what keeps and passes on the news)
and client (what people use to read news) for IBM mainframes.  I was one
of the first people to beta-test it, and I was also a subscriber to several
LISTSERV mailing lists.  I knew that Usenet servers exchange their news as
(essentially) e-mail, so I asked my friend to add a feature to NETNEWS so
that it could exchange e-mail with a LISTSERV mailing list:  a gateway.
I knew about the BIOSCI mailing lists (there were only mailing lists then),
but at the time I was under the impression that subscribers had to have
accounts on the computer administered by BIOSCI at that time (most did).

While my friend was off conducting a discussion about this in Usenet, Dave 
Kristofferson at BIOSCI and his colleagues set up their own gateway system,
using an independently written (but identically named) program.  For various,
now irrelevant reasons, that arrangement did not work very smoothly and it
was eventually dismantled (in 1990?) in favor of the system we now have,
where the bionet groups have two custom gateways into mailing lists, one
on net.bio.net and the other on daresbury.ac.uk.  Meanwhile, my friend did
create a gateway for SAS-L at UGA, my favorite list, into bit.listserv.sas-l.
That gateway has been going fine, with no more bounces than we see in the
bionet groups, since 1987. 

So the LISTSERV <=> Usenet gateway does work.  The question is:  do we want
it to work in bionet?  I say yes:  it would foster a sense of community
and make life simpler for many administrators who now keep track of the
alt.*, bit.listserv.*, and sci.bio.* hierarchies as well as bionet.*, just
to follow all the biology research groups.

Actually, there's no reason why people can't integrate the two systems
locally.  Using the "tin" package, as Tony Travis in bionet.plants has
often said, it is easy to set up a subscription to any mailing list and
have it distributed "downstream" as a local Usenet group.  But it would
be nice if everyone had that.

-- 

      Una Smith      Biology Department       smith-una at yale.edu
                     Yale University
                     New Haven, CT  06511