github.com/arnoldrobbins.qed-archive
1
0
Fork 0
mirror of synced 2026-01-11 23:43:06 +00:00
Code Issues Releases Wiki Activity

Added some QED on Multics history.

Browse Source
This commit is contained in:
Arnold D. Robbins 2024-10-31 19:13:31 +02:00
parent a4ef5cf22c
commit 7d85818d82
2 changed files with 145 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
multics/README.md
View File

@ -10,6 +10,9 @@ These files were supplied to me by Charles Anthony
There are source files and documentation files. I have no idea
how to build or use them. Enjoy.
The file `some-history.txt` is from email received in October, 2024.
The contents are included here by permission.
## Other Links
Here is [a cheat sheet for Multics QED](http://www.bitsavers.org/pdf//honeywell/multics/swenson/6906.multics-condensed-guide.pdf).
@ -17,4 +20,4 @@ Thanks to Doug McIlroy.
#### Last Updated
Mon Feb 1 07:54:43 IST 2021
Thu Oct 31 19:12:45 IST 2024

141
multics/some-history.txt Normal file
View File

@ -0,0 +1,141 @@
Thu Oct 31 07:06:52 PM IST 2024
===============================
The following text was forwarded to me by Clem Cole. It's included
here by permission from Paul and Tom.
Arnold Robbins
-------------------------------------------
From Paul Green, October 29, 2024:
Hi Clem (and Tom),
Clem, you asked whether the source code for “QED”, as archived at
https://github.com/arnoldrobbins/qed-archive, accurately represents
Multics source code. Also, by implication, whether this is an
implementation of QED or some closely-related editor.
The first question is easy to answer. Yes, this is Multics source code.
I know of Charles Anthony, and I’m pretty sure that I know the origins of
this source code. Tom Van Vleck (cc’d) performed the herculean task of
prying the source code for Multics out of Bull in France, who ended up as
the last holder of the rights to the code. Tom and I both worked on Multics
for Honeywell Information Systems in Cambridge (MA) and our names or
initials are all over the source code. Please visit
https://web.mit.edu/multics-history/ for more information on the Multics
source code.
Charles Anthony and several others wrote an emulator for the Multics
hardware which has been operational for several years, and that team
continues to enhance and maintain a version of the Multics code for that
environment. I’m a user of that emulation environment, working on
resurrecting a program for a friend. That code is also on GitHub somewhere.
Answering the second question requires some background. I probably don’t
know the whole story here (e.g., I don’t know the story of QED on CTSS) and
that’s why I’ve copied Tom. Here’s what I do know. I distinctly recall
Robert F. Mabee (Bob) maintaining a version of QED written in BCPL. This
would be in the ~1969-1971 timeframe. I don’t think he wrote it. I don’t
know who wrote it, and would be surprised if its source code exists
anywhere. I don’t think many people used it. It had a reputation of being
slow and heavy on system resources (CPU and paging). QED was also, at least
to my college-student eyes, quite baroque in its command set and thus hard
to learn. BCPL was also not a popular programming language on Multics, so
few people could fix problems in it or enhance it. Thankfully, Jerry
Saltzer wrote edm (‘editor for Multics’) as a greatly simplified line
editor which was much simpler to learn and use, and much leaner on its
consumption of system resources. In those days, Multics used printing
terminals at 150-300 baud, and supported a few dozen users in about a
megabyte of core memory. Compilations of large (several thousand+ lines)
programs could take many minutes because a CPU heavy user got less CPU time
than an interactive user. I don’t recall who wrote qedx as a way of taking
the power of QED and the efficiency and simplicity of edm, but they did a
great job. Qedx quickly became the editor of choice, and stayed that way
until we got display terminals and emacs (written in Lisp, naturally).
So the code in this GitHub archive should probably be called “inspired
by QED”. I don’t think it is fair to say that it *is* QED. My direct
knowledge of Multics spans the 1969 to mid-1980. I can’t speak to changes
to Multics after I left in April, 1980.
In the late 60s and well into the 70s, a printing terminal cost as much
as a VW bug ($2K). So we didn’t have personal terminals; we had a pool room
where we shared access to a hodge-podge of various models of printing
terminals. Smoking was still a thing, and I tended to avoid the times when
a smoker was present in the terminal room. Honeywell had two terminal
rooms, and management eventually decreed that one of them was no-smoking,
which was wonderful. Toward the end of the 1970s, the whole office became
non-smoking, over many objections.
PG
---------------------------------------------------
From Tom Van Vleck, October 30, 2024:
The article on QED in https://multicians.org/mgq.html covers the CTSS
version of QED written by Ken Thompson in about 1965, and its descendants.
In the 1965-55 time frame, Ken worked for Bell Labs in Murray Hill NJ
as part of the Multics development team.
To use Ken's CTSS QED, one invoked the command on a BCD file, and then
issued commands to insert or delete lines, or alter the "current line."
This QED editor provided a command that let the user specify a text
sequence with a "regular expression" that could match part of a line,
and a "substitute" command that could replace the text matching a regular
expression with new text.
Multics development began at MIT Project MAC in 1965, using CTSS as
a tool. One major change from CTSS to Multics was that Multics used the
ASCII character set, which had upper and lower case letters and more
punctuation marks than the BCD character set used on CTSS. CTSS QED
was enhanced to read ASCII files as well as BCD. Many of the Multics
developers used QED to write PL/I source.
Ken Thompson wrote a version of his CTSS QED in the BCPL language. (Martin
Richards visited MIT Project MAC in the mid 60s and ported the Cambridge
BCPL language to CTSS.) Bell Labs programmers ported BCPL and the
BCPL QED to the early Multics as Multics became self-hosting in 1968.
Bell Labs programmers also wrote roff in bcpl and ported it to Multics.
As Paul mentions, the early versions of Multics had performance problems,
and using the BCPL version of QED used a lot of resources. In a brilliant
solo development, Bob Daley created the QEDX editor for Multics: it was
a simplification of QED with probably about 80% of the editing commands
and much better performance. Multics developers migrated over to qedx
for most usage.. a few editing features that were omitted from qedx
were needed by the Multics documentation group, so the BCPL qed was not
deleted, but was used only in rare cases.
(Bob Daley was one of the principal developers of CTSS, and then became
the de-facto "chief engineer" for Multics in the 60s. He decided what
to work on next, what features were in and what were not, who should do
what, how to trade function for performance.)
Ken's CTSS version of QED was inspired by his
previous usage of Butler Lampson's Berkeley QED. See
https://en.wikipedia.org/wiki/QED_(text_editor) but CTSS and Multics I/O
was line-at-a-time as opposed to character-at-a-time in the Berkeley
system and the DEC PDP computers. The line orientation was derived
from the way that timesharing terminal I/O was implemented, and was not
changed until the late 70s when Multics support for emacs was created.
On CTSS in the early 60s, the earliest editors, INPUT and EDIT, imitated
card punching. Users printed out the whole file with line numbers, and
then edited the file by replacing whole lines. INPUT and EDIT worked on
CTSS files in the BCD character set, which had uppercase letters only.
One breakthrough on CTSS was the 1963 "context editor" CTEST9, created
by Art Samuel of IBM, which allowed a user to "peek" at the file contents
while changing it.
Jerry Saltzer wrote a context editor called TYPSET for CTSS in 1964.
It edited files in the "12-bit" character set, which supported upper and
lower case letters. See https://en.wikipedia.org/wiki/TYPSET_and_RUNOFF
Later, Jerry modified the character I/O routines for TYPSET to create EDM,
which edited ASCII files.
(Dan Murphy had created the TECO editor for editing paper tape on the
PDP-1 in 1962. It had little influence on text editing on CTSS and
Multics, but was influential on DEC computers and the MIT AI Lab's
ITS system.)
Hope this helps.