>>....    Thus,  your REXX  list  would  be  comp.lang.rexx, ....
>While this seems like a good idea at first glance, you cannot start giving
>these groups standard Usenet names without first getting approval by the
>Usenet hierarchy. Creating a group comp.lang.rexx would almost definitely
>require sending postings from it into Usenet, ...
 
In general this is true. "comp.", "talk.", etc. should be used for
approved Usenet groups only. But "comp.lang.rexx is a bad example,
because it already really exists :-)  It is the REXXLIST, gatewayed at
Berkeley into Usenet. BTW, it is a bit messy. I assume the gateway isn't
subscribed with the fullheader option, therefore the Message-Id is lost
and Berkeley has to create a new one. Whenever a redistribution list
or some kind of gateway is subscribed to Listserv, the subscription
should be accompanied by a SET FULL command. If this is done properly,
multiple gateway may feed the wonderful world of Rexx programming into
Usenet to enlighten all these Unix-C-hackers.
 
>I therefore favor using bit. names. We get a feed from Penn State, and
>Bill Verity calls the groups 'bit.listserv.groupname' where 'groupname'
>is the actual LISTSERV name of the group. This way has always seemed
>fine to me, as it makes it simple for users to map the Netnews group
>to the corresponding LISTSERV group.
 
Except for some Bitnet groups that have really Usent counterparts
(the example above, INFO-M2 = comp.lang.modula2, etc.), we need
- an appropriate prefix (examples: "bit.", "bit.listserv.",
  "bitmail.", ...)
- an ending that corresponds to the Bitnet list name.
 
The prefix should be short, and indicate it's origin. "bit." doesn't
indicate that these are gatewayed mailing lists. "bit.listserv."
is a bit long. "bitmail." matches all requirements. This is of
course my personal biased opinion, and you need no guessing to find
out what I'm using here :-)  But anything else would be ok with me
(as long as it has less than 30 characters).
 
The ending should be the original list name, even if this is ugly.
It's bad enough that there are different list names on Arpa and Bitnet,
to add a third naming variant would only create more confusion (if
this is possible).
 
Result: FOOBAR-L  -->  bit.listserv.foobar-l  or  bitmail.foobar-l
(BTW, please note that this isn't a bidirectional arrow!)
 
Maybe we should continue this discussion on NETNWS-L@NDSUVM1 only.
 
  Thomas