From: Russ Allbery (rra@stanford.edu)
Date: Sun Feb 01 2004 - 13:53:37 CST
Bruce Lilly <blilly@erols.com> writes:
> If, having once relayed an article, a server's "backlog" indicates that
> it yet has to relay that same article, the server's software is indeed
> broken.
Agreed.
> If that variety of brokenness is commonplace,
It is.
> then we should address that issue somewhere (not in the article format).
It already is addressed, and addressing this issue is vital to the
robustness of the netnews protocol. You have to address this issue in
order to permit a flood-fill algorithm for propagating messages, and the
combination of the history database and rejecting articles with too old of
a Date does exactly this.
This MUST NOT be compromised by any revision of the Usenet protocols. It
is nearly as integral to the proper functioning of Usenet as the presence
of the Message-ID header.
-- Russ Allbery (rra@stanford.edu) <http://www.eyrie.org/~eagle/>