Xref: utzoo comp.mail.uucp:1546 news.config:816 comp.mail.misc:1147
Path: utzoo!utgpu!water!watmath!clyde!mcdchg!heiby
From: heiby@mcdchg.UUCP (Ron Heiby)
Newsgroups: comp.mail.uucp,news.config,comp.mail.misc,chi.mail
Subject: Re: ihnp4 cutoff
Message-ID: <11147@mcdchg.UUCP>
Date: 4 Aug 88 14:46:59 GMT
References: <122@att.ATT.COM> <3400001@eecs.nwu.edu>
Reply-To: heiby@mcdchg.UUCP (Ron Heiby)
Organization: Motorola Microcomputer, Schaumburg, IL
Lines: 17

Jacob Gore (gore@eecs.nwu.edu) writes:
> All neighbors of the att triad must declare it as att(DEAD).

I don't understand why this is the case.  If the links from att to the
outside world (like mcdchg) aren't published anywhere, pathalias should
not be generating any paths that contain the string "att!mcdchg".  Of
course, the link from att to mcdchg should be published *within* AT&T,
so pathalias runs for AT&T systems *can* show that link.

My understanding of pathalias is that a link decalared from A -> B
*does not* imply a link from B -> A at better than DEAD cost, anyway.
So, it's true that if the ONLY CONCEIVABLE WAY to get to a site is
from B -> A at some point, pathalias will generate such a path, but
it knows that the "cost" is astronomical.
-- 
Ron Heiby, heiby@mcdchg.UUCP	Moderator: comp.newprod & comp.unix
"Failure is one of the basic Freedoms!" The Doctor (in Robots of Death)