[Simh] SDS 940 TSS bring-up update

Mon Mar 2 18:29:42 EST 2015

All,

Background: In 2005 Al Kossow supplied me with several thousand SDS 940
files extracted from archival tapes. I attempted to bring a system up using
Bob Supniks excellent SDS 940 simulator, but became frustrated with what
appeared to be several missing pieces of the monitor. After failing to
locate additional 940 tapes in the basement of the NBS Radio Building in
Boulder where we used to store them in the 1970s, I put the project aside,
being more than full-time employed then.

In 2010 I retired and moved to Mexico. I did not pursue the 940 project
until a former NBS colleague, Howard Bussey, broached the subject in
January, 2014, with suggestions of alternative approaches. In turn, that
prodded me to write a number of tools to scan the file collection looking
for needed components of the Monitor and Executive, and to create a
simulated drum.  Debugging and development has continued intermittently for
the last 14 months, resulting in a state of work that perhaps merits wider
discussion of where to go with it, and whether anyone else wants to join my
efforts.

At the end of this e-mail are my notes on the current project status.  In
summary, I have a nice running SDS 940 time sharing system with components
to edit, assemble and link code.  In the eating your own dog food
department, I can assemble, link and run the system and make changes to it;
no cross-assembly necessary. I have no doubt that if I installed the
simulator on a Raspberry Pi attached to my router, any of you could telnet
in and have a productive session.

Theres still a lot to do in terms of finding and building the remaining
subsystems like FORTRAN, BASIC and SNOBOL. Getting the system built with a
permanent hard disk file system instead of the drums temporary file system
would make this easier, not having to copy files from the host each session
via simulated paper tape.

Does anyone think theres sufficient interest to make this a GITHUB
collaborative project? Or should I just labor away as time and energy
permits?  For privacy reasons, I cant make the multi-thousand file archive
public, because theres personal information such as resumes buried among
the archives that must be searched for relevant source files.  However, I
could put on GITHUB just those files Ive found as well as the tools Ive
developed. The effort to manage a GITHUB repository would only be worth it
if others contributed. As a distribution system to just a few folks, I can
do that more easily as a zip file.

Many thanks to Al Kossow for the original tape files, to Bob Supnik for an
excellent simulator, to Mark Pizzolato for his patience with my many
screwups sending bugfixes to the SIMH archive, and to Howard Bussey for
poking me to give this another try.

Mark Emmer

Pátzcuaro, Michoacán, México

---------------------------------------

Received about 2,470 extracted files from the L16x tape set.  These included
source, binary, save and dump format files. Wrote a number of C tools to
find needed code amongst the memory save files and to convert to formats
needed by the simulator.

By finding memory save image files of needed components, was able to build a
simulated drum with the Time Sharing System's Monitor, Exec, user directory
and these functional applications:

NARP - New ARPAS Assembler

DDT - Linker, loader, debugger

QED - Text editor

CAL - Conversational Algebraic Language

LISP - LISP programming language

CCP - Conditional Command Processor

Can transfer files between the host O/S and 940 exec via paper tape reader
and punch. Files are stored on drum. Multiple simultaneous users can be
accommodated via multiple telnet sessions.

CCP program that stuffs input to and steals output from a pseudo teletype to
process command files works properly.

Can reassemble and link the Monitor and Exec and install on drum using a CCP
command file.  Have created a paper-tape bootstrap to load the production or
debug monitor from the drum based on console breakpoint switch setting.

Fixed various errors in the 940 simulator that caused Monitor, Exec and
subsystem malfunctions.

To Do:

1.	Rebuild Monitor and Exec to include support for Data Products or
Bryant movable arm disk to provide permanent file storage.

2.	Attempt to build all applications from source files.  In addition to
those listed above, need to find source files to build: OLDDDT, SNOBOL3,
SNOBOL4, QSPL, RUNOFF, BASIC, FORTRAN II, FORTRAN IV and MAIL.  Will add
them to the Exec command tables to make them built-in commands, rather than
applications that are launched by the GO TO <savefile> command.

3.	Find the OPER program so can manage users and file directories.

4.	Find the NEWTAPE program to properly initialize a mag tape with a
file directory.

5.	Get the Monitor and Exec assembly symbols that are saved on drum to
work properly with DDT so can debug/inspect a running or crashed system. The
mechanism used was designed for OLDDDT, which so far I have not been able to
locate.

6.	Having the QSPL compile running will allow investigation of the many
QSPL source programs in the collection.

7.	Comb through the 2,470 files as well as files on the L14x tape set
for other interesting programs.  For example, L14x appears to contain files
for PCAP, an early class scheduling program, with data files that make it
look like it was used by the E.E. department at UCB.
8.	Create a front panel display for the running system.

Failures:

1.	Never could build a proper mag tape image that would work with the
system.  The system build was for a RAD/tape system with no disk storage.
This SDS release uses an overwrite-in-place mechanism to maintain a
random-access read/write file system on tape, complete with user
directories. Never did find the NEWTAPE program that's supposed to
initialize a tape, and all attempts to create a binary tape image that would
satisfy the Exec's tape mount command failed. While I can boot a system from
a mag tape image file, all file transfers between the host and the TSS have
to be accomplished via simulated paper tape.

Blue-Sky for someone else To Do:

1.	There's a build switch in the monitor for SRI, and tantalizing hints
of code for Douglas Englebart's NLS, such as finding that the SMT (Shared
Memory Table) reserves 8 drum pages for NLS. Theres 940 code in the monitor
to interface to a PDP-5. However, to date, no source code has been found for
NLS. To duplicate the original SDS 940 hardware implementation would require
a PDP-5 simulator with associated CRT display and information on how it was
attached to the SDS 940.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.trailing-edge.com/pipermail/simh/attachments/20150302/2400b50d/attachment.html>