Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10 UW 5/3/83; site uw-beaver
Path: utzoo!watmath!clyde!burl!ulysses!mhuxl!houxm!houxz!vax135!cornell!uw-beaver!info-mac
From: info-mac@uw-beaver (info-mac)
Newsgroups: fa.info-mac
Subject: Re: Instant Pascal
Message-ID: <1326@uw-beaver>
Date: Thu, 26-Jul-84 12:20:32 EDT
Article-I.D.: uw-beaver>.1326
Posted: Thu Jul 26 12:20:32 1984
Date-Received: Fri, 27-Jul-84 07:27:05 EDT
Sender: daemon@uw-beave
Organization: U of Washington Computer Science
Lines: 30

From: Scott M. Hinnrichs <SMH@SRI-KL.ARPA>
	You may want to do some more research on what the tty drivers
and other system services are able to do as far as blocked and character
data.  It is easily possible to make a IBM type interface for a UNIX
system where it is not as easily possible to go the other way.  Just ask
AMDAHL.  Admittedly it is the hardware philosophy getting in the way,
but nevertheless it is an inherent problem when dealing with IBM shops.
	You are on the right track though.  We are trying to establish,
through the use of a standard terminal line, a dialogue which can occur
between an intelligent user interface such as the MAC and any mainframe.
The goal is to make a definition of the protocol for MAINFRAME <-> MAC
handshaking, take the idea from there and implement an application on
top of that link.
	One possibility is to take the idea of sockets/ports and treat
the connection like a network protocol using the line as the base
medium.
	First, construct a command protocol to run on one stream as in
a remote-procedure calling mechanism.  This can be made to be flexible
enough to provide basic communication requests, and yet provide the
needed hooks to quote more application specific requests.
	Second, provide the data channel needed to transmit the data
requests made in the command stream.  There are some considerations in
this that would suggest that it should use some standard graphics
protocol other than what is currently available.  Initially it can
merely move the data in checksummed datagrams.
	This is to be implemented soon, so if anyone has any ideas I
would appreciate hearing from you now.

Scott
-------