Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!uunet!wang!fitz
From: fitz@wang.UUCP (Tom Fitzgerald)
Newsgroups: news.config
Subject: Re: deleted UUCP sites
Message-ID: <777@wang.UUCP>
Date: 3 Jan 90 19:56:26 GMT
References: <4195@convex.UUCP> <30273@mcdchg.chg.mcd.mot.com> <9575@samsung.samsung.com>
Distribution: news
Organization: Wang Labs, Lowell MA
Lines: 22

warren@samsung.COM (Warren Lavallee) writes:
> I think the UUCP map coordinators should require that the node exists,
> before allowing an edge in a map entry.  Also, every so often they
> should do consistency checks on all of their maps and take corrective
> measures if they do not pass.

I have to agree with this (which is a little hypocritical because there
is a machine here at Wang that violates this, but what the heck).  When
people publish connections to unmapped machines, it really raises the
chance of getting 2 machines with the same name in different places
(which has already happened with 'castle' and with 'alice' I believe).
This leads down the path of confused aggressive-rerouters, which can
only end in chaos and madness.

That doesn't mean the map coordinators should do a surprise-rewrite on
map entries, though.  Maybe when a site is going to get expired, its
neighbors should get notified before the map entry gets trashed?

---
Tom Fitzgerald   fitz@wang.com          It's a mistake to believe that
Wang Labs        ...!uunet!wang!fitz    the die is cast.  In reality, it
Lowell MA, USA   1-508-967-5278         is injection-molded.