🚀 go-pugleaf

Path: sun!gorodish!guy
From: guy%gorodish@Sun.COM (Guy Harris)
Newsgroups: mod.protocols
Subject: Re: Submission for mod-protocols
Message-ID: <13682@sun.uucp>
Date: 20 Feb 87 05:35:58 GMT
References: <8702191514.AA20193@bunny.UUCP>
Sender: news@sun.uucp
Reply-To: guy@sun.UUCP (Guy Harris)
Organization: Sun Microsystems, Mountain View
Lines: 40

>Courier and Clearinghouse were some of the finest higher level
>protocols to exist when they were published. Bill Joy of Sun Microsystems
>liked them so much, he rewrote them for TCP/IP and called them RPC and
>YellowPages, respectfully.

Well, aside from the fact that:

	1) Bill didn't write them single-handedly (for example,
	the manager of the NFS group here, and one of the developers
	of Sun RPC and RPC-based services, is Bob Lyon - yes, the same
	Bob Lyon listed as one of the authors of the clearinghouse at
	the end of Oppen and Dalal's TOOIS paper).

	2) They weren't "rewritten for TCP/IP"; Sun RPC has an
	object-oriented interface to the underlying transport protocol
	(you provide an interface module to the transport in question -
	which need not be a network protocol, but could be
	something like shared memory - that provides a certain set of
	operations) and has been put on top of, e.g., the ISO connectionless
	transport protocol (and could probably be put on top of, e.g.,
	SPP).  Also, Courier performs the functions of RPC and XDR
	combined, not just RPC.

	3) The Yellow Pages is a lower-level remote database access
	protocol than Clearinghouse.  The ranges of the mappings it
	performs consist of uninterpreted data, not property lists.
	It does not support queries like Clearinghouse's
	"LookupGeneric", because it doesn't interpret those data; you
	can look up something by name, or you can get every entry in
	a map and do whatever kind of lookup you want, but you can't
	say "give me all the objects that match this wildcarded name
	and that have thus-and-such a property."  It also has no
	mechanism for updating mappings.

I guess this isn't *too* far off (other than the fact that the proper
term is "respectively", not "respectfully").

If you only have a few widely-spaced data points, be careful when
attempting interpolation; it looks pretty silly when you try to use
linear interpolation on data points taken from a sine wave....