Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!uunet!wuarchive!psuvax1!psuvm!TUBVM!HABERNOL
From: HABERNOL@TUBVM.CS.TU-BERLIN.DE (Thomas Habernoll)
Newsgroups: bit.listserv.policy-l
Subject: Re: Technical direction vs. Services
Message-ID: <POLICY-L%90021100252299@DEARN>
Date: 10 Feb 90 22:43:48 GMT
Sender: Discussion about BITNET policies <POLICY-L@BITNIC>
Reply-To: Discussion about BITNET policies <POLICY-L@BITNIC>
Lines: 32
Approved: NETNEWS@PSUVM Gateway

>>                                                         However, like
>>LISTSERV and NETSERV, NETNEWS does not seem to encourage imitators.
>>There is no specifications document that would allow  enterprising
>>programmers on other systems to develop compatible programs.
>Sorry, but here I disagree totally.  There is a specifications document,
>and it's RFC's 977 and 1036.  Taken together, they should describe enough
>to allow someone who wanted to write a NETNEWs-like beast for their
>environment to create something which could communicate with the rest of
>the Usenet world.

Nope. Maybe the RFC's _should_ describe enough, but they _don't_.
You can't write a Netnews system from scratch by only looking at
the RFC. For the nitty gritty details you have to look how other
systems are doing the work (and even then there are still enough
controversy parts).

This is not said to diminish the worth of Netnews, I'm preaching
Netnews for years now. And even a weak RFC is more than we have for
other critical network servers.

And Listserv? Well, feel free to design, implement, and document! a
better Listserv. Nobody would complain. But make sure that it provides
such an effective transport mechanism as the current Listerv is
doing. This is for most users the invisible part (contrary to
mailing lists and file server functions), but it's the most important
one in my opinion. Not just for mail/news, but for the distribution
of software, too. There are _large_ software packages that can be sent
to 10, 20, or even 100+ sites by Listserv's DISTribute mechanism
while putting no more load on many links than a single copy would do.
If this only wouldn't depend so much on an underlying NJE network ...

  Thomas