joel@gould9.UUCP (Joel West) (01/31/86)
Although I have previously fallen for the "domainist" cult, I'm
beginning to rethink my views.
Ironically, this comes at the hands of Eric Allman, the original
sendmail warrior. He is the author of "Interview with Peter
Honeyman", in UNIX Review, January 1986, p. 64.
This is the first clear, relatively cogent exposition of Honeyman's
beliefs I've seen. Me thinks he's rushing things a bit when he
posts his Nth follow-up on domain-less addressing to net.mail.
While domainism is convenient in theory, Honeyman notes that
In a name space that tries to support absolute names,
like the ARPANET or the Internet, authority resides
in the hands of a few. But in UUCP, authority
resides in the hands of anybody who can edit a
system's L.sys file.
I, for one, think I know how to make relative names
work, and I'm not in the least bit pessimistic about
my ability to keep scaling up as the network grows.
One of the interesting things about the UUCP network, with
its thousands and thousands of hosts, is that no
host is more than seven hops away from any other.
It's a pretty well connected network.
(I don't know enough topology to understand his later comments about
the "sparse" UUCP network graph.)
The article also talks about network standards. I agree (more or less)
with Honeyman, uucp is only a de facto standard for the UNIX systems;
in more heterogeneous environments, other protocols are appropriate.
If uucp were to become a successful heterogenous standard, I think it
would have to compete with standards that have been placed in the public
domain and under the control of standards organizations.
Joel West (619) 457-9681
CACI, Inc. Federal, 3344 N. Torrey Pines Ct., La Jolla, CA 92037
{cbosgd,ihnp4,pyramid,sdcsvax,ucla-cs}!gould9!joel
gould9!joel@nosc.ARPA
PS: In case you didn't gather from my previous excerpt from this issue,
this UNIX REVIEW is very good for those interested in networking.