NetNews Usenet Archive 1992 #16/NN_1992_16.iso/spool/comp/arch/8273

Text File | 1992-07-24 | 5.2 KB | 135 lines

Newsgroups: comp.arch
Path: sparky!uunet!decwrl!deccrl!news.crl.dec.com!stewart
From: [e-mail] (Larry Stewart)
Subject: Proposal: Computer History Project
Message-ID: <[e-mail]>
Keywords: software, history, computer
Sender: [e-mail] (USENET News System)
Reply-To: [e-mail]
Organization: DEC Cambridge Research Lab
Date: Fri, 24 Jul 1992 14:46:43 GMT
Lines: 122

Here is the beginning of an idea.  Let's kick it around and build on it.
I've been thinking about this for 6 months or so, and I've mentioned
the idea to several people privately, but klh's recent announcement
galvanized me into action.
    -Larry Stewart


THE COMPUTER HISTORY PROJECT

Lawrence C. Stewart
Cambridge Research Laboratory
Digital Equipment Corporation
July 24, 1992

Summary
-------

The idea of the Computer History Project is to preserve the
history of computing systems, and to make that history readily
available to everyone.  The basic idea is to publish a CDROM,
or perhaps a series of CDROMs which would contain emulators
for historic computing machines and copies of their operating
software and documentation.

Modern computers are so much faster than historic machines that
it is possible to emulate the instruction set and I/O systems
of a historic machine at full speed.  Thus students will be
able to actually sit down and use TENEX running on an emulated
PDP-10, or Bravo on a Xerox Alto, even when no more PDP-10's or
Altos exist.

The history of computing is partly a story of hardware, and
that is well captured by film and museums.  However, the
largest part of computing is software, and at present there is
no means of preserving it.

Preserving Software
-------------------

There are three parts to the problem of preserving software:

1)    The software is kept on older media types, for which
    I/O devices are not readily available.

2)    The hardware on which the software runs is obsolete
    and no longer maintainable.

3)    The documentation for the software and systems is frequently
    on paper.

The media problem must be handled by rolling forward old software
onto new media.  At the moment, the CDROM represents the best technology
for preserving older software - it is cheap, it is read-only, it
has high capacity, and the I/O devices will be widely available for
some time to come.

The hardware problem must be handled by emulation.  We have the unique
advantage in computing that machine performance is constantly increasing.
Views vary, but it seems that performance at constant cost is increasing
at about 50% to 100% per year.  The industry has been on this curve for
a long time and there seems no obstacle in view that will prevent it
from continuing.  Consequently, software that emulates an older machine
can do so at "full performance" and indeed it should be possible to build
"faster" machines in software than was possible when they were embodied
in hardware.

The problem of documentation is the most difficult, but the proposal
is that the critial documentation for using historic software be
scanned and stored in electronic form as compressed images.  A CDROM
can store a great many pages this way.  It may also be possible to use
optical character recognition to eventually convert old paper documentation
into computer readable text.

Emulating the Software
----------------------

It will not be sufficient to produce and distribute emulators just for
the instruction set of a historic computer.  Much of what makes these
systems interesting is the user environment and the system software.  For
this reason, the emulators will have to either emulate the operating system
interfaces or emuate the I/O devices and actually run a copy of the original
operating system as well as the applications.

This may represent a lot of work, but it is by no means impossible.
For example, Ken Harrenstein has recently announced an emulator for the
DEC KS-10 processor sufficiently accurate that the MIT ITS system can be
run.

How will the project work?
--------------------------

There must be several components:

A set of guidelines for emulation and portability of the emulators.
The emulators will have to be included in source form, so that the
emulators can be rolled forward.

A clearinghouse to collect electronic contributions to the project.
This should be a computer system accessible to the Internet, so that
systems submitted to the clearinghouse can be exposed to the "beta test"
of the internet community before they are frozen on a CDROM.

A set of guidelines for the formats of stored data and documentation must
be created.  There are a variety of compressed image storage formats
available, for example.

In the case of commercial systems, the vendors must release
into the public domain copies of operating systems, utility software,
and documentation.  At the minimum, vendors must be encouraged
to preserve complete copies of software and documentation until
such time as they are prepared to release them into the public
domain.

Conclusion
----------

This is just a sketch of an idea.  In less than one lifetime the computing
industry has progressed from nothing to todays 100 MIP desktop machines.
We must not let the opportunity slip away to preserve the history of
this phenomenon.

======================================================================