Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!rutgers!cs.utexas.edu!csd4.milw.wisc.edu!lll-winken!arisia!sgi!shinobu!odin!fudge!karlton
From: karlton@fudge.sgi.com (Phil Karlton)
Newsgroups: comp.windows.x
Subject: Re: Can X Be Used As A General-Purpose IPC Mechanism?
Message-ID: <171@odin.SGI.COM>
Date: 28 Jun 89 18:59:32 GMT
References: <19833@cup.portal.com>
Sender: news@odin.SGI.COM
Organization: Silicon Graphics, Inc., Mountain View, CA
Lines: 30

In article <19833@cup.portal.com> Will@cup.portal.com (Will E Estes) writes:
>Can X be used as a general-purpose IPC mechanism across a network, or
>is X so specifically tailored to user-interface functions that it is
>not well-suited towards more general IPC and client/server functions?

While X could be used to implement an IPC mechanism, it would not be
very efficient. On page xxiii of the Digital Press _X Window System_ book,
there is a paragraph that addresses this point. The last sentence is

    The X protocol is correctly viewed as just one component in an
    overall distributed system architecture, not as the complete
    architecture by itself.

There is only one client/server function that X was intentionally
designed to support.

Let me encourage somebody out there to design, implement and give away an
OS and transport independent RPC mechanism. You could become famous.

>The way I understand it, X officially supports TCP/IP and DECnet.
>Are there plans to support TOPS, Novell, the Microsoft LAN Manager,
>and Appletalk?  (Did I leave out any other obvious nets?)

X doesn't actually support any transport mechanism; it's more the
other way around.  The X protocol can be built on top of any reliable
byte stream. As you note, there already exist implementations on top
of TCP/IP and DECnet. ChaosNet seems like an obvious example of a
network that somebody should be working on.

PK