ietf-nntp Dropping the Distribution option from Newnews?

Brian Kantor brian at karoshi.ucsd.edu
Fri Feb 20 10:12:59 PST 1998


>Is Mr Kantor still around to elaborate on what was originally intended
for this feature?

Yup, I'm here.

The intention was that the distribution specification in the NEWNEWS
command could be used to filter articles presented in the newnews output,
restricted by the contents of the Distribution: header in the articles.
I.e., NEWNEWS would only present a host with articles that matched the news
groups and distributions specified.

Remember that Distribution: was designed to restrict the propagation of
news - so that someone could, e.g., post to the newsgroup "net.forsale"
(which had implied worldwide coverage) yet set the Distribution header to
"soca", which would restrict the article to NOT being delivered to news
servers not serving Southern Californa.

With the advent of non-regional news service providers, such as Netcom,
AOL, etc., the concept of restricting article FLOW by geographic region is
pretty much moot.  However, retrieval is a different matter - it still
makes sense to me to be able to post and retrieve articles which were
intended for my home town of San Diego, even if I'm getting them from a
server in Boston or Podunk.

If properly used, Distribution headers (and the corresponding clause in
the NEWNEWS command) could eliminate the need for regional newsgroups.

I wouldn't hold my breath waiting for THAT to happen.

Because of some bad design decisions made in earlier popular news servers
(which were default installed fecklessly passing all distributions to each
news neighbor), distributions are mostly useless and have been largly
superceded by regional newsgroup hierarchies.

A last note: please to remember that the NNTP RFC (977) was written 11
years ago, BEFORE we had much experience in shipping and retrieving news
with the protocol.  It's only now with Stan et al that there will be a new
RFC that reflects the reality of current practice.  So if something isn't
compliant with RFC977, don't condem the code - it might well be an
improvement.  (What do you expect from a protocol designed by a couple of
undergrads?)  Internet Lesson#1: standardise working code, not design ideas.
	- Brian




More information about the ietf-nntp mailing list