Discussion:
Unix vs. OS/360 (history)
(too old to reply)
nt_mahmood@yahoo.com [hercules-390]
2016-04-20 15:29:53 UTC
Permalink
Hello,
I asked a question on stackexchange about a historical thing then I remembered that once I was working with Hercules, I asked so many questions here and received very valuable replies.

So, I thought it is better to share my question here. Excuse me if it is unrelated to Hecules itself...

I read the history of Unix operating system [1] and also read the original Unix paper [2] by Thompson and Ritchie (the pdf is available at berkeley's website [3]).



In their paper they mention some key features which Unix uses. Some terminologies including file handling (read, write, ...), process management (fork, ...), user access controls (super user, permissions, ...), were defined and explained in that paper.



My question is, were those words and terminologies really new at that time (1974)?



If the answer is yes, then how were mainframes working before Unix? I mean how IBM OS/360 https://en.wikipedia.org/wiki/OS/360_and_successors was working? Didn't they (IBM guys) use file and processes for batch jobs and storing information?Didn't they use text editors for their programs?


If the answer is no, then what was really new in Unix?!



Regards,
Mahmood

[1] https://en.wikipedia.org/wiki/History_of_Unix https://en.wikipedia.org/wiki/History_of_Unix
[2] http://dl.acm.org/citation.cfm?id=361061 http://dl.acm.org/citation.cfm?id=361061
[3] http://www.cs.berkeley.edu/~brewer/cs262/unix.pdf http://www.cs.berkeley.edu/~brewer/cs262/unix.pdf
'Dave G4UGM' dave.g4ugm@gmail.com [hercules-390]
2016-04-20 15:58:14 UTC
Permalink
I think many of these concepts came from Multics and were not in common use elsewhere. In particular, the Mainframe in 1974 had no concept of “Fork”, “Super User” etc. Not sure when user permissions appeared but OS/VS2 aka MVS had only just been release in 1974 and I don’t think MVT or MFT had permissions, or even users. The concept of “Users” first appeared in TSO for MVT but there were no file permissions as such. Passwords could be assigned to files as per:-



http://bitsavers.informatik.uni-stuttgart.de/pdf/ibm/360/os/tso/GC28-6698-5_Time_Sharing_Option_Guide_Jul72.pdf



but I am fairly sure these would have no effect in batch.



Dave

G4UGM



From: hercules-***@yahoogroups.com [mailto:hercules-***@yahoogroups.com]
Sent: 20 April 2016 16:30
To: hercules-***@yahoogroups.com
Subject: [hercules-390] Unix vs. OS/360 (history)





Hello,
I asked a question on stackexchange about a historical thing then I remembered that once I was working with Hercules, I asked so many questions here and received very valuable replies.

So, I thought it is better to share my question here. Excuse me if it is unrelated to Hecules itself...

I read the history of Unix operating system [1] and also read the original Unix paper [2] by Thompson and Ritchie (the pdf is available at berkeley's website [3]).



In their paper they mention some key features which Unix uses. Some terminologies including file handling (read, write, ...), process management (fork, ...), user access controls (super user, permissions, ...), were defined and explained in that paper.



My question is, were those words and terminologies really new at that time (1974)?



If the answer is yes, then how were mainframes working before Unix? I mean how IBM OS/360 <https://en.wikipedia.org/wiki/OS/360_and_successors> was working? Didn't they (IBM guys) use file and processes for batch jobs and storing information?Didn't they use text editors for their programs?



If the answer is no, then what was really new in Unix?!



Regards,
Mahmood

[1] https://en.wikipedia.org/wiki/History_of_Unix
[2] http://dl.acm.org/citation.cfm?id=361061
[3] http://www.cs.berkeley.edu/~brewer/cs262/unix.pdf
'John P. Hartmann' jphartmann@gmail.com [hercules-390]
2016-04-20 16:25:46 UTC
Permalink
I beg to disagree. At the time of UNIX's inception (1970) OS/MVT was
widespread.

There was nothing new in UNIX, except that it was designed by a few
sharp minds rather than a committee, so the design was elegant and
compact. fork() was a convenient result of the swapping rather than
paging implementation on the original hardware.

Organic even pointed out that Multics had pipes, but that nobody had
figured out how to use them.

-------- Forwarded Message --------
Subject: RE: [hercules-390] Unix vs. OS/360 (history)
Date: Wed, 20 Apr 2016 16:58:14 +0100
From: 'Dave G4UGM' ***@gmail.com [hercules-390]
<hercules-***@yahoogroups.com>
Reply-To: hercules-***@yahoogroups.com
To: hercules-***@yahoogroups.com



I think many of these concepts came from Multics and were not in common
use elsewhere. In particular, the Mainframe in 1974 had no concept of
“Fork”, “Super User” etc. Not sure when user permissions appeared but
OS/VS2 aka MVS had only just been release in 1974 and I don’t think MVT
or MFT had permissions, or even users. The concept of “Users” first
appeared in TSO for MVT but there were no file permissions as such.
Passwords could be assigned to files as per:-

http://bitsavers.informatik.uni-stuttgart.de/pdf/ibm/360/os/tso/GC28-6698-5_Time_Sharing_Option_Guide_Jul72.pdf

but I am fairly sure these would have no effect in batch.

Dave

G4UGM

*From:*hercules-***@yahoogroups.com [mailto:hercules-***@yahoogroups.com]
*Sent:* 20 April 2016 16:30
*To:* hercules-***@yahoogroups.com
*Subject:* [hercules-390] Unix vs. OS/360 (history)



Hello,
I asked a question on stackexchange about a historical thing then I
remembered that once I was working with Hercules, I asked so many
questions here and received very valuable replies.

So, I thought it is better to share my question here. Excuse me if it is
unrelated to Hecules itself...

I read the history of Unix operating system [1] and also read the
original Unix paper [2] by Thompson and Ritchie (the pdf is available at
berkeley's website [3]).

In their paper they mention some key features which Unix uses. Some
terminologies including file handling (read, write, ...), process
management (fork, ...), user access controls (super user, permissions,
...), were defined and explained in that paper.

My question is, were those words and terminologies really new at that
time (1974)?

If the answer is yes, then how were mainframes working before Unix? I
mean how IBM OS/360
<https://en.wikipedia.org/wiki/OS/360_and_successors> was working?
Didn't they (IBM guys) use file and processes for batch jobs and storing
information?Didn't they use text editors for their programs?

If the answer is no, then what was really new in Unix?!



Regards,
Mahmood

[1] https://en.wikipedia.org/wiki/History_of_Unix
[2] http://dl.acm.org/citation.cfm?id=361061
[3] http://www.cs.berkeley.edu/~brewer/cs262/unix.pdf
'Dave Wade' dave.g4ugm@gmail.com [hercules-390]
2016-04-20 16:58:40 UTC
Permalink
-----Original Message-----
Sent: 20 April 2016 17:26
Subject: Fwd: RE: [hercules-390] Unix vs. OS/360 (history)
I beg to disagree. At the time of UNIX's inception (1970) OS/MVT was
widespread.
I never said it wasn't. What I said was it doesn't have Super User, Fork or file level permissions...
There was nothing new in UNIX, except that it was designed by a few sharp
minds rather than a committee, so the design was elegant and compact. fork()
was a convenient result of the swapping rather than paging implementation on
the original hardware.
Organic even pointed out that Multics had pipes, but that nobody had figured
out how to use them.
And Multics not IBM Mainframes were the primary sources of input to UNIX.
In fact the differences between UNIX and Mainframe paradigms are about as different as one can get.

So one can postulate that having evolved from "unit record" or punched card/tabulators then fixed length records was an obvious way for mainframes to develop.
GE/Honeywell GCOS has similar file constructs, but lacks the variable block size of CKD devices.

On the other hand, starting with a Teletype for i/o then perhaps it was natural for Multics and UNIX to adopt the stream model for data.

When it comes to the underlying structures I believe that IBM tended to adopted fixed length tables with fixed size blocks with rigid structures. Things like file names tend to be embedded within the control block. Contrast that with UNIX, and I am not sure if it was in the original design, but many of the structures have pointers to strings or arrays, and the whole feel is very different.

Dave
-------- Forwarded Message --------
Subject: RE: [hercules-390] Unix vs. OS/360 (history)
Date: Wed, 20 Apr 2016 16:58:14 +0100
I think many of these concepts came from Multics and were not in common use
elsewhere. In particular, the Mainframe in 1974 had no concept of “Fork”,
“Super User” etc. Not sure when user permissions appeared but
OS/VS2 aka MVS had only just been release in 1974 and I don’t think MVT or
MFT had permissions, or even users. The concept of “Users” first appeared in
TSO for MVT but there were no file permissions as such.
Passwords could be assigned to files as per:-
http://bitsavers.informatik.uni-stuttgart.de/pdf/ibm/360/os/tso/GC28-6698-
5_Time_Sharing_Option_Guide_Jul72.pdf
but I am fairly sure these would have no effect in batch.
Dave
G4UGM
*Sent:* 20 April 2016 16:30
*Subject:* [hercules-390] Unix vs. OS/360 (history)
Hello,
I asked a question on stackexchange about a historical thing then I
remembered that once I was working with Hercules, I asked so many questions
here and received very valuable replies.
So, I thought it is better to share my question here. Excuse me if it is unrelated
to Hecules itself...
I read the history of Unix operating system [1] and also read the original Unix
paper [2] by Thompson and Ritchie (the pdf is available at berkeley's website
[3]).
In their paper they mention some key features which Unix uses. Some
terminologies including file handling (read, write, ...), process management
(fork, ...), user access controls (super user, permissions, ...), were defined and
explained in that paper.
My question is, were those words and terminologies really new at that time (1974)?
If the answer is yes, then how were mainframes working before Unix? I mean
how IBM OS/360 <https://en.wikipedia.org/wiki/OS/360_and_successors> was
working?
Didn't they (IBM guys) use file and processes for batch jobs and storing
information?Didn't they use text editors for their programs?
If the answer is no, then what was really new in Unix?!
Regards,
Mahmood
[1] https://en.wikipedia.org/wiki/History_of_Unix
[2] http://dl.acm.org/citation.cfm?id=361061
[3] http://www.cs.berkeley.edu/~brewer/cs262/unix.pdf
------------------------------------
------------------------------------
http://groups.yahoo.com/group/hercules-390
http://www.hercules-390.org
------------------------------------
Yahoo Groups Links
'John P. Hartmann' jphartmann@gmail.com [hercules-390]
2016-04-20 17:23:38 UTC
Permalink
Dave, we are way off topics here, but the reason for the unlimited
string length in UNIX was the null terminated rather than length,data
paradigm of OS.

Which flowed into Windows (the C string paradigm, that is).

What would you say about an MVS system that has as many security flaws
as Windows has/had due to that one maldesign?
Post by 'Dave Wade' ***@gmail.com [hercules-390]
Contrast that with UNIX, and I am not sure if it was in the original
design, but many of the structures have pointers to strings or arrays,
and the whole feel is very different.
'Dave Wade' dave.g4ugm@gmail.com [hercules-390]
2016-04-20 17:50:22 UTC
Permalink
-----Original Message-----
Sent: 20 April 2016 18:24
Subject: Re: [hercules-390] Unix vs. OS/360 (history)
Dave, we are way off topics here, but the reason for the unlimited string length
in UNIX was the null terminated rather than length,data paradigm of OS.
Is this from the OS or from the language chosen to write the code in, originally assembler, then "B" and the "C"....
Which flowed into Windows (the C string paradigm, that is).
What would you say about an MVS system that has as many security flaws as
Windows has/had due to that one maldesign?
That’s a totally different topic and totally off-topic. The question was "what things did Unix inherit from Mainframes" and in the context of IBM mainframes the answer is "virtually nothing".

Dave,
390]
Post by 'Dave Wade' ***@gmail.com [hercules-390]
Contrast that with UNIX, and I am not sure if it was in the original
design, but many of the structures have pointers to strings or arrays,
and the whole feel is very different.
------------------------------------
------------------------------------
http://groups.yahoo.com/group/hercules-390
http://www.hercules-390.org
------------------------------------
Yahoo Groups Links
Bob Flanders bob.flanders@gmail.com [hercules-390]
2016-04-20 17:51:20 UTC
Permalink
http://www-03.ibm.com/systems/z/os/zos/features/racf/

IBM Introduced RACF in 1974, but if I understand correctly, it's optional.
Post by 'Dave Wade' ***@gmail.com [hercules-390]
-----Original Message-----
Sent: 20 April 2016 17:26
Subject: Fwd: RE: [hercules-390] Unix vs. OS/360 (history)
I beg to disagree. At the time of UNIX's inception (1970) OS/MVT was
widespread.
I never said it wasn't. What I said was it doesn't have Super User, Fork
or file level permissions...
There was nothing new in UNIX, except that it was designed by a few sharp
minds rather than a committee, so the design was elegant and compact.
fork()
was a convenient result of the swapping rather than paging
implementation on
the original hardware.
Organic even pointed out that Multics had pipes, but that nobody had
figured
out how to use them.
And Multics not IBM Mainframes were the primary sources of input to UNIX.
In fact the differences between UNIX and Mainframe paradigms are about as
different as one can get.
So one can postulate that having evolved from "unit record" or punched
card/tabulators then fixed length records was an obvious way for mainframes
to develop.
GE/Honeywell GCOS has similar file constructs, but lacks the variable
block size of CKD devices.
On the other hand, starting with a Teletype for i/o then perhaps it was
natural for Multics and UNIX to adopt the stream model for data.
When it comes to the underlying structures I believe that IBM tended to
adopted fixed length tables with fixed size blocks with rigid structures.
Things like file names tend to be embedded within the control block.
Contrast that with UNIX, and I am not sure if it was in the original
design, but many of the structures have pointers to strings or arrays, and
the whole feel is very different.
Dave
-------- Forwarded Message --------
Subject: RE: [hercules-390] Unix vs. OS/360 (history)
Date: Wed, 20 Apr 2016 16:58:14 +0100
I think many of these concepts came from Multics and were not in common
use
elsewhere. In particular, the Mainframe in 1974 had no concept of “Fork”,
“Super User” etc. Not sure when user permissions appeared but
OS/VS2 aka MVS had only just been release in 1974 and I don’t think MVT
or
MFT had permissions, or even users. The concept of “Users” first
appeared in
TSO for MVT but there were no file permissions as such.
Passwords could be assigned to files as per:-
http://bitsavers.informatik.uni-stuttgart.de/pdf/ibm/360/os/tso/GC28-6698-
5_Time_Sharing_Option_Guide_Jul72.pdf
but I am fairly sure these would have no effect in batch.
Dave
G4UGM
*Sent:* 20 April 2016 16:30
*Subject:* [hercules-390] Unix vs. OS/360 (history)
Hello,
I asked a question on stackexchange about a historical thing then I
remembered that once I was working with Hercules, I asked so many
questions
here and received very valuable replies.
So, I thought it is better to share my question here. Excuse me if it is
unrelated
to Hecules itself...
I read the history of Unix operating system [1] and also read the
original Unix
paper [2] by Thompson and Ritchie (the pdf is available at berkeley's
website
[3]).
In their paper they mention some key features which Unix uses. Some
terminologies including file handling (read, write, ...), process
management
(fork, ...), user access controls (super user, permissions, ...), were
defined and
explained in that paper.
My question is, were those words and terminologies really new at that
time
(1974)?
If the answer is yes, then how were mainframes working before Unix? I
mean
how IBM OS/360 <https://en.wikipedia.org/wiki/OS/360_and_successors> was
working?
Didn't they (IBM guys) use file and processes for batch jobs and storing
information?Didn't they use text editors for their programs?
If the answer is no, then what was really new in Unix?!
Regards,
Mahmood
[1] https://en.wikipedia.org/wiki/History_of_Unix
[2] http://dl.acm.org/citation.cfm?id=361061
[3] http://www.cs.berkeley.edu/~brewer/cs262/unix.pdf
------------------------------------
------------------------------------
http://groups.yahoo.com/group/hercules-390
http://www.hercules-390.org
------------------------------------
Yahoo Groups Links
Tony Harminc tharminc@gmail.com [hercules-390]
2016-04-20 16:34:47 UTC
Permalink
Post by 'Dave G4UGM' ***@gmail.com [hercules-390]
I think many of these concepts came from Multics and were not in common
use elsewhere. In particular, the Mainframe in 1974 had no concept of
“Fork”, “Super User” etc.
True, but MVT (and even later MFT) had the notion of ATTACH, which is
arguably more general than fork().
Post by 'Dave G4UGM' ***@gmail.com [hercules-390]
Not sure when user permissions appeared but OS/VS2 aka MVS had only just
been release in 1974
1972, I think. Though maybe not really in customer land until early 1974.

and I don’t think MVT or MFT had permissions, or even users. The concept of
Post by 'Dave G4UGM' ***@gmail.com [hercules-390]
“Users” first appeared in TSO
Well... There were plenty of multi-user application programs already
around, e.g. APL\360, CRJE, ATS, CPS from IBM, and Wylbur and others from
non-IBM sources. These all had their own user/password databases, not known
to the OS itself. TSO was somewhat more integrated into the OS (MVT 20.0)
than other such systems, but still the user/password database wasn't used
for anything else, and there was no batch TSO in the early days.

for MVT but there were no file permissions as such. Passwords could be
Post by 'Dave G4UGM' ***@gmail.com [hercules-390]
assigned to files as per:-
http://bitsavers.informatik.uni-stuttgart.de/pdf/ibm/360/os/tso/GC28-6698-5_Time_Sharing_Option_Guide_Jul72.pdf
but I am fairly sure these would have no effect in batch.
Au contraire, dataset (loosely "file") passwords worked (and still do in
z/OS, I believe) in batch. The operator was prompted for the password,
which isn't a good match for today's ideas on security, separation of
roles, and so on. The passwords are stored in a dataset called "PASSWORD"
(which is itself password protected, of course).

At the University of Toronto where I worked in the early 1970s, a couple of
academics had actually written a research paper and implemented a small mod
so that an authorized (loosely, since APF didn't exist in MVT) program
could place a magic string + password at a fixed offset from an OPEN SVC,
and it would be used without the annoying operator prompt. (One of them may
even have been Dave Wortman, who was one of the authors of XPL, now being
discussed in another thread.)

Tony H.
'Dave Wade' dave.g4ugm@gmail.com [hercules-390]
2016-04-20 17:41:31 UTC
Permalink
-----Original Message-----
Sent: 20 April 2016 18:11
Subject: Re: [hercules-390] Unix vs. OS/360 (history)
Post by 'Dave G4UGM' ***@gmail.com [hercules-390]
I think many of these concepts came from Multics and were not in
common use elsewhere. In particular, the Mainframe in 1974 had no
concept of "Fork", "Super User" etc. Not sure when user permissions
appeared but
OS/VS2 aka MVS had only just been release in 1974 and I don't think
MVT or MFT had permissions, or even users. The concept of "Users"
first appeared in TSO for MVT but there were no file permissions as
such.
Post by 'Dave G4UGM' ***@gmail.com [hercules-390]
Passwords could be assigned to files as per:-
I agree about Multics, but I think that the Burroughs large systems had
files etc.
in the mid 60s- even if the filesystem was flat (i.e. no datasets, no
directories).
A job could definitely spawn child processes via the "cactus stack"
approach,
but the semantics were very different from those of fork().
Does this give you a copy of the current process. Most of the other tools
such as "attach" allow you to run a program as a child, but, it's a
different program, and if you do re-run the same program you don't get
access to its variable space, open handles as you do with fork() etc. see

http://linux.die.net/man/2/fork

it really is a unique way for working...
However Burroughs very much had their finger on the academic pulse at the
time, and I strongly suspect that at least some of this stuff came via a
(Californian?) university.
I think many of the BUNCH had different paradigms to IBM
--
Mark Morgan Lloyd
markMLl .AT. telemetry.co .DOT. uk
[Opinions above are the author's, not those of his employers or
colleagues]
Dave
'Fernando M. Roxo da Motta' mvs@roxo.org [hercules-390]
2016-04-20 18:43:06 UTC
Permalink
On Wed, 20 Apr 2016 17:11:24 +0000, "Mark Morgan Lloyd
Post by 'Dave G4UGM' ***@gmail.com [hercules-390]
I think many of these concepts came from Multics and were not in
common use elsewhere. In particular, the Mainframe in 1974 had no
concept of “Fork”, “Super User” etc. Not sure when user permissions
appeared but OS/VS2 aka MVS had only just been release in 1974 and
I don’t think MVT or MFT had permissions, or even users. The
concept of “Users” first appeared in TSO for MVT but there were no
file permissions as such. Passwords could be assigned to files as
per:-
I agree about Multics, but I think that the Burroughs large systems
had files etc. in the mid 60s- even if the filesystem was flat (i.e.
no datasets, no directories). A job could definitely spawn child
There is a description of Burroughs operating system MCP at Wikipedia:

https://en.wikipedia.org/wiki/Burroughs_MCP
processes via the "cactus stack" approach, but the semantics were
very different from those of fork().
However Burroughs very much had their finger on the academic pulse at
the time, and I strongly suspect that at least some of this stuff
came via a (Californian?) university.
Roxo

--
---------------- Non luctari, ludare -------------------+ WYSIWYG
Fernando M. Roxo da Motta <***@roxo.org> | Editor?
Except where explicitly stated I speak on my own behalf.| VI !!
( Usuário Linux registrado #39505 ) | I see text,
------------ Quis custodiet ipsos custodes?-------------+ I get text!
Mahmood Naderan nt_mahmood@yahoo.com [hercules-390]
2016-04-21 05:51:58 UTC
Permalink
Thank you guys for the feedback. Here are some notes about some of the replies
Post by Harold Grovesteen ***@tx.rr.com [hercules-390]
For mainframes, time-sharing usually
coexisted in a system designed for batch jobs.
That isn't to say that there were no exclusively
time-sharing systems that ran on the mainframe
of the day. VM/CMS and MTS were. But
these were rarer than your TSO
sitting amidst a bunch of batch jobs.
Sorry can you explain more since I have not worked with mainframes. What I read from the history of mainframes is that users carry some punch cards, give them to the operator and waits waits for their turn. So that is really time sharing... Sharing some resources in time. Isn't that?
Then what was the purpose of Multics? Bell Labs, MIT and GE were trying to design a tiem sharing system!
Do you mean that IBM was on business a didn't pay attention to others, so they tried to build every thing from scratch? Similar thing happened to the Unix itself. When every thing was licensed by Bell Labs, R. Stallman started to write a free compiler from scratch but that was really existing on the market
Post by Harold Grovesteen ***@tx.rr.com [hercules-390]
The critical difference is in file organization.
Unix is basically byte oriented, and
distinguishes binary and text files. IBM's files
are record based (e.g., a punch card or print
line), that made it much more efficient than Unix
access. With some exceptions a program or>utility could operate on a file without knowing
the contents type.
For me it is weird how it is possible to work on a file without accessing its content. Do you have any example for that? Again, for me which have no experience with old systems, such things are strange but I really like to know about that. According to the Unix paper, they also defined some i-nodes (I think similar to records) to put some properties along with the content.
Post by Harold Grovesteen ***@tx.rr.com [hercules-390]
Obviously, the name, supposedly derived from Multics.
Unix provided decent computing on small machines,
with some requisite features (permissions, etc.),
but the basic concepts weren't really new.
Thanks for that. I do agree with that.  
Post by Harold Grovesteen ***@tx.rr.com [hercules-390]
That’s a totally different topic and totally off-topic.
The question was "what things did Unix inherit
from Mainframes" and in the context of IBM
mainframes the answer is "virtually nothing".
Well I think Ritchie and Thompson were agree with you because they didn't refer to any mainframe concept in their paper! But from the above, I actually think they tried to *port* some big features that was mostly applicable to big systems, e.g. mainframes, to mini computers for personal computers.
Do you agree with that?


Regards,
Mahmood
'John P. Hartmann' jphartmann@gmail.com [hercules-390]
2016-04-21 08:10:34 UTC
Permalink
Wikipedia offers opinion, not facts.

On 04/21/2016 09:40 AM, Mark Morgan Lloyd
*** Most of this stuff is summarised adequately on Wikipedia. ***
------------------------------------

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

Community email addresses:
Post message: hercules-***@yahoogroups.com
Subscribe: hercules-390-***@yahoogroups.com
Unsubscribe: hercules-390-***@yahoogroups.com
List owner: hercules-390-***@yahoogroups.com

Files and archives at:
http://groups.yahoo.com/group/hercules-390

Get the latest version of Hercules from:
http://www.hercules-390.org


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

Yahoo Groups Links

<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/hercules-390/

<*> Your email settings:
Individual Email | Traditional

<*> To change settings online go to:
http://groups.yahoo.com/group/hercules-390/join
(Yahoo! ID required)

<*> To change settings via email:
hercules-390-***@yahoogroups.com
hercules-390-***@yahoogroups.com

<*> To unsubscribe from this group, send an email to:
hercules-390-***@yahoogroups.com

<*> Your use of Yahoo Groups is subject to:
https://info.yahoo.com/legal/us/yahoo/utos/terms/
Mahmood Naderan nt_mahmood@yahoo.com [hercules-390]
2016-04-21 08:14:30 UTC
Permalink
*** Most of this stuff is summarised adequately on Wikipedia. ***
Yes... at the time i was reading that I got confused.. Thanks for pointing out that.. I will read it again
 Regards,
Mahmood
'Dave Wade' dave.g4ugm@gmail.com [hercules-390]
2016-04-21 09:16:05 UTC
Permalink
That's actually true of many encyclopaedias. Wikipedia generally has a long
list of supporting references for each article, many with links so it is
often possible to assess the information for oneself.

Ascertaining influences is almost always impossible, even if you can speak
to the original author, as they many actually have forgotten how or where
the learnt the idea from, or more simply they may not want to disclose it so
they can claim ownership. I don't believe this is the case with Thompson and
Ritchie. They seem to delight in telling us what their influences were, and
Mainframes were what they were escaping from. So whilst it might have had an
input, I am pretty sure it was negative, i.e. how do we avoid all that
baggage such as JCL, Blocking, etc. and lets have something as simple as we
can...


Dave
-----Original Message-----
Sent: 21 April 2016 09:11
Subject: Re: [hercules-390] Unix vs. OS/360 (history)
Wikipedia offers opinion, not facts.
On 04/21/2016 09:40 AM, Mark Morgan Lloyd markmll.hercules-
*** Most of this stuff is summarised adequately on Wikipedia. ***
------------------------------------
------------------------------------
http://groups.yahoo.com/group/hercules-390
http://www.hercules-390.org
------------------------------------
Yahoo Groups Links
'Dave Wade' dave.g4ugm@gmail.com [hercules-390]
2016-04-21 10:52:04 UTC
Permalink
Agreed, but at the same time there are certain things which are so basic
that
nobody is really sure where they originated. As an example, what
architecture
introduced the combined parameter/return stack which is close to being a
prerequisite of C? It's something that just about everybody takes for
granted,
and I'd certainly not like to try to cite anybody for its invention.
Well "B" had the same stack/structure a "C" as did I think BCPL. Doesn't ir
really go back to Turing and the Turing Machine and isn't the tape in effect
an infinite length stack?
--
Mark Morgan Lloyd
markMLl .AT. telemetry.co .DOT. uk
Dave
G4UGM
'Dave Wade' dave.g4ugm@gmail.com [hercules-390]
2016-04-21 13:42:04 UTC
Permalink
Post by 'Dave Wade' ***@gmail.com [hercules-390]
Well "B" had the same stack/structure a "C" as did I think BCPL.
Doesn't ir really go back to Turing and the Turing Machine and isn't
the tape in effect an infinite length stack?
But I don't think there's provision for saving the state of the state
machine in
the reader/writer to the tape.
My suspicion would be something like the IAS machine, I don't know whether
e.g. the Manchester "Baby" was a contender on account of its extremely
limited memory. A useful approach would probably be to find what machine
introduced the idea of an interrupt which implicitly saved at least some
registers, e.g. due to a non-present fault on paged or segmented memory.
The "Baby" really is the minimal implementation required to write a program.
So only 7 instructions, not even an add, and no concept of a stack. However,
it illustrates the problem with deducing influences very well.

When I am demonstrating the replica, I am frequently asked "what did Turing
have to do with the machine" and generally as far as we can tell nothing, he
didn't come to Manchester till after baby was working.
However there is a huge desire of folks to link Turing in. They can't
conceive of a computer being built in the UK without him. They point out
that bits of on Colossus from Bletchly came to Manchester, were they used in
Baby. Here we know the answer is "No". Kilburn looked at the circuits, which
were I believe a ring counter and discarded them as being totally
unsuitable.

So deducing influences is hard.
--
Mark Morgan Lloyd
markMLl .AT. telemetry.co .DOT. uk
[Opinions above are the author's, not those of his employers or
colleagues]
------------------------------------
------------------------------------
http://groups.yahoo.com/group/hercules-390
http://www.hercules-390.org
------------------------------------
Yahoo Groups Links
myyahoo@chycoski.com [hercules-390]
2016-04-22 22:50:33 UTC
Permalink
When I started working at a university in 1975, we had an OS/MVT system that had userids and passwords, and project numbers that were related to filesystem access control and privileged program access. We also had WYLBUR/MILTEN for editing and batch job submission and retrieval (we did not have ORVYL), and APL/SV for online timesharing. WYLBUR had concepts of privilege for filesystem access and execution of certain commands - and 'underprivileged' which prevented a user from changing their password (yeah, something you wouldn't see much of today! :-).

Our environment was highly customised, so some of these features were peculiar to our site.
Harold Grovesteen h.grovsteen@tx.rr.com [hercules-390]
2016-04-20 16:25:40 UTC
Permalink
The Ritchie paper really answers your question in Section 8.1,
Influences. No they were not new.

However, UNIX came out of the fairly new, at the time, world of
minicomputers. Mainframes already had a decade of history at that
point. And computing in general had another decade or so of history
before mainframes. Mainframes at this time were mostly used for batch
processing. Whereas, Multics (Bell Labs/MIT), a major influence on
UNIX, was a time-sharing system. For mainframes, time-sharing usually
coexisted in a system designed for batch jobs. That isn't to say that
there were no exclusively time-sharing systems that ran on the mainframe
of the day. VM/CMS and MTS were. But these were rarer than your TSO
sitting amidst a bunch of batch jobs.

These major differences in system size, foundational technology and use
resulted in different approaches to the running software. The whole
UNIX style of computing came out of the research and educational
communities. The mainframe style of computing came out of use by the
business community to run their businesses. These two styles of
computing have only really recently come together with Linux running on
mainframes. Until then, these worlds have been separate due to these
differences in their respective roots.

Harold Grovesteen
Post by ***@yahoo.com [hercules-390]
Hello,
I asked a question on stackexchange about a historical thing then I
remembered that once I was working with Hercules, I asked so many
questions here and received very valuable replies.
So, I thought it is better to share my question here. Excuse me if it
is unrelated to Hecules itself...
I read the history of Unix operating system [1] and also read the
original Unix paper [2] by Thompson and Ritchie (the pdf is available
at berkeley's website [3]).
In their paper they mention some key features which Unix uses. Some
terminologies including file handling (read, write, ...), process
management (fork, ...), user access controls (super user,
permissions, ...), were defined and explained in that paper.
My question is, were those words and terminologies really new at that time (1974)?
If the answer is yes, then how were mainframes working before Unix? I
mean how IBM OS/360 was working? Didn't they (IBM guys) use file and
processes for batch jobs and storing information?Didn't they use text
editors for their programs?
If the answer is no, then what was really new in Unix?!
Regards,
Mahmood
[1] https://en.wikipedia.org/wiki/History_of_Unix
[2] http://dl.acm.org/citation.cfm?id=361061
[3] http://www.cs.berkeley.edu/~brewer/cs262/unix.pdf
'Thomas Valerio' tjv@westwood-tech.com [hercules-390]
2016-04-20 17:19:38 UTC
Permalink
That isn't to say that there were no exclusively time-sharing systems
that ran on the mainframe of the day. VM/CMS and MTS were.
If by "exclusively time-sharing systems" you mean NO batch capability, I'm
not sure where this came, I can't comment on VM/CMS, but MTS was
absolutely NOT an "exclusively time-sharing system". Although it had the
word "Terminal" in it's name, and the focus was on time-sharing and
terminal based access it offered batch capability from the get-go.

Thomas Valerio
The Ritchie paper really answers your question in Section 8.1,
Influences. No they were not new.
However, UNIX came out of the fairly new, at the time, world of
minicomputers. Mainframes already had a decade of history at that
point. And computing in general had another decade or so of history
before mainframes. Mainframes at this time were mostly used for batch
processing. Whereas, Multics (Bell Labs/MIT), a major influence on
UNIX, was a time-sharing system. For mainframes, time-sharing usually
coexisted in a system designed for batch jobs. That isn't to say that
there were no exclusively time-sharing systems that ran on the mainframe
of the day. VM/CMS and MTS were. But these were rarer than your TSO
sitting amidst a bunch of batch jobs.
These major differences in system size, foundational technology and use
resulted in different approaches to the running software. The whole
UNIX style of computing came out of the research and educational
communities. The mainframe style of computing came out of use by the
business community to run their businesses. These two styles of
computing have only really recently come together with Linux running on
mainframes. Until then, these worlds have been separate due to these
differences in their respective roots.
Harold Grovesteen
Post by ***@yahoo.com [hercules-390]
Hello,
I asked a question on stackexchange about a historical thing then I
remembered that once I was working with Hercules, I asked so many
questions here and received very valuable replies.
So, I thought it is better to share my question here. Excuse me if it
is unrelated to Hecules itself...
I read the history of Unix operating system [1] and also read the
original Unix paper [2] by Thompson and Ritchie (the pdf is available
at berkeley's website [3]).
In their paper they mention some key features which Unix uses. Some
terminologies including file handling (read, write, ...), process
management (fork, ...), user access controls (super user,
permissions, ...), were defined and explained in that paper.
My question is, were those words and terminologies really new at that time (1974)?
If the answer is yes, then how were mainframes working before Unix? I
mean how IBM OS/360 was working? Didn't they (IBM guys) use file and
processes for batch jobs and storing information?Didn't they use text
editors for their programs?
If the answer is no, then what was really new in Unix?!
Regards,
Mahmood
[1] https://en.wikipedia.org/wiki/History_of_Unix
[2] http://dl.acm.org/citation.cfm?id=361061
[3] http://www.cs.berkeley.edu/~brewer/cs262/unix.pdf
'John P. Hartmann' jphartmann@gmail.com [hercules-390]
2016-04-20 17:25:25 UTC
Permalink
Count APL\360 in.
That isn't to say that there were no exclusively time-sharing systems
that ran on the mainframe of the day. VM/CMS and MTS were.
------------------------------------

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

Community email addresses:
Post message: hercules-***@yahoogroups.com
Subscribe: hercules-390-***@yahoogroups.com
Unsubscribe: hercules-390-***@yahoogroups.com
List owner: hercules-390-***@yahoogroups.com

Files and archives at:
http://groups.yahoo.com/group/hercules-390

Get the latest version of Hercules from:
http://www.hercules-390.org


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

Yahoo Groups Links

<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/hercules-390/

<*> Your email settings:
Individual Email | Traditional

<*> To change settings online go to:
http://groups.yahoo.com/group/hercules-390/join
(Yahoo! ID required)

<*> To change settings via email:
hercules-390-***@yahoogroups.com
hercules-390-***@yahoogroups.com

<*> To unsubscribe from this group, send an email to:
hercules-390-***@yahoogroups.com

<*> Your use of Yahoo Groups is subject to:
https://info.yahoo.com/legal/us/yahoo/utos/terms/
'Dave Wade' dave.g4ugm@gmail.com [hercules-390]
2016-04-20 17:44:55 UTC
Permalink
-----Original Message-----
Sent: 20 April 2016 18:20
Subject: Re: [hercules-390] Unix vs. OS/360 (history)
That isn't to say that there were no exclusively time-sharing systems
that ran on the mainframe of the day. VM/CMS and MTS were.
If by "exclusively time-sharing systems" you mean NO batch capability, I'm
not
sure where this came, I can't comment on VM/CMS, but MTS was absolutely
NOT an "exclusively time-sharing system". Although it had the word
"Terminal"
in it's name, and the focus was on time-sharing and terminal based access
it
offered batch capability from the get-go.
I would say that whilst MTS and VM/CMS have a batch capability, unlike the
OS family there is no separate JCL commands and commands regardless of
environment....
Thomas Valerio
Dave
Gerhard Postpischil gerhardp@charter.net [hercules-390]
2016-04-20 16:22:22 UTC
Permalink
Post by ***@yahoo.com [hercules-390]
In their paper they mention some key features which Unix uses. Some
terminologies including file handling (read, write, ...), process
management (fork, ...), user access controls (super user, permissions,
...), were defined and explained in that paper.
My question is, were those words and terminologies really new at that time (1974)?
There was, and still is, a disconnect between the academic and
commercial worlds. IBM pretty much did its own thing, and ignored
everything else. Their machines were mostly dedicated to large data
centers, and nothing like permissions existed (prior to RACF, a data set
could be protected, but an operator had to supply the password, via
console, for each access). The fork process is completely different -
IBM has an ATTACH command that creates a subtask, but it lacked some of
the Unix features.
Post by ***@yahoo.com [hercules-390]
If the answer is yes, then how were mainframes working before Unix? I
mean how IBM OS/360
<https://en.wikipedia.org/wiki/OS/360_and_successors> was working?
Didn't they (IBM guys) use file and processes for batch jobs and storing
information?Didn't they use text editors for their programs?
Early IBM systems didn't require an editor. Users carried around trays
of punch cards, and edited those. The 709x systems had a primitive
update capability (a predecessor to IEBUPDAT). More usable editors came
in around the seventies (TSO EDIT, Wylbur, etc.). The critical
difference is in file organization. Unix is basically byte oriented, and
distinguishes binary and text files. IBM's files are record based (e.g.,
a punch card or print line), that made it much more efficient than Unix
access. With some exceptions a program or utility could operate on a
file without knowing the contents type.

IBM competitors (also referred to as the seven dwarves) more or less
used similar approaches. RCA dropped out, and that left the BUNCH
(Burroughs, Univac, NCR, CDC, Honeywell).

There was a second level tier of companies that produced smaller
machines - "mini-computers" (e.g., Digital Equipment Corporation - DEC),
but essentially every manufacturer had a distinct operating system, with
non-standard extensions to ANSI standard compilers. IBM more or less set
the bar, and competitors has to match features. With a few exceptions,
academic institutions used smaller machines, and provided systems
designed to keep concurrent users from interfering with each other,
Post by ***@yahoo.com [hercules-390]
If the answer is no, then what was really new in Unix?!
Obviously, the name, supposedly derived from Multics. Unix provided
decent computing on small machines, with some requisite features
(permissions, etc.), but the basic concepts weren't really new.


Gerhard Postpischil
Bradford, VT
Peter Rosenberg peter_rosenberg_dk@yahoo.co.uk [hercules-390]
2016-04-21 08:35:07 UTC
Permalink
The concept of users must have been there before TSO came around. I believe it came when (IBM) system CICS were developed for (warehouse I believe) and also IMS were in its infancy, before MVS was there.Then, online terminals were interacting with applications that were transaction driven. Hence a scheduling system was needed to trigger the applications that should pull the transaction from the queue, process it and possibly deliver a feedback to the end user.Teller machines, came around too, and I cant imagine any of these pre-TSO system were working without a User concept, since accountability (and Traceability etc) were a must.I had the pleasure of moving an IMS security based User Management system, to RACF back in the mid-eighties, and this was a left over from a VM/CMS variant were customers IMS DB/DC system ran.
Hope it explains. And sorry I haven't had the time to pick source evidence for above reply.
Peter Rosenberg
'Dave Wade' dave.g4ugm@gmail.com [hercules-390]
2016-04-21 09:17:08 UTC
Permalink
Ealy CICS used fixed terminal IDS


From: hercules-***@yahoogroups.com [mailto:hercules-***@yahoogroups.com]
Sent: 21 April 2016 09:35
To: hercules-***@yahoogroups.com
Subject: [hercules-390] Re: Unix vs. OS/360 (history)





The concept of users must have been there before TSO came around. I believe it came when (IBM) system CICS were developed for (warehouse I believe) and also IMS were in its infancy, before MVS was there.
Then, online terminals were interacting with applications that were transaction driven. Hence a scheduling system was needed to trigger the applications that should pull the transaction from the queue, process it and possibly deliver a feedback to the end user.
Teller machines, came around too, and I cant imagine any of these pre-TSO system were working without a User concept, since accountability (and Traceability etc) were a must.
I had the pleasure of moving an IMS security based User Management system, to RACF back in the mid-eighties, and this was a left over from a VM/CMS variant were customers IMS DB/DC system ran.

Hope it explains. And sorry I haven't had the time to pick source evidence for above reply.

Peter Rosenberg
stephen.orso@yahoo.com [hercules-390]
2016-04-21 16:45:48 UTC
Permalink
Early versions of CICS did support userids and passwords, statically defined in DFHSNT macros and used by the CSSN/CSSF transactions. However, significant improvement would have been needed to make that capability rise to the level of something that would be considered primitive by today's understanding.

Applications that required security usually had it baked into the application, either by checking terminal ids and letting the application run only from selected terminals, using the CSSN id and maintaining a separate table or file of users and permissions, or just implementing an application security system not unlike Oracle's E-Business Suite does today.


The two companies that I worked for that used CICS used them for customer order entry and fulfillment, and for spare parts inventory.


IBM eventually dumped CSSN/CSSF and expected third-party solutions (CA Top-Secret for example) to fill the void. The IBM support site mentions a utility to migrate DFHSNT entries into RACF as an alternative for OS users.


---In hercules-***@yahoogroups.com, <***@...> wrote :

Ealy CICS used fixed terminal IDS


From: hercules-***@yahoogroups.com [mailto:hercules-***@yahoogroups.com]
Sent: 21 April 2016 09:35
To: hercules-***@yahoogroups.com
Subject: [hercules-390] Re: Unix vs. OS/360 (history)







The concept of users must have been there before TSO came around. I believe it came when (IBM) system CICS were developed for (warehouse I believe) and also IMS were in its infancy, before MVS was there.

Then, online terminals were interacting with applications that were transaction driven. Hence a scheduling system was needed to trigger the applications that should pull the transaction from the queue, process it and possibly deliver a feedback to the end user.

Teller machines, came around too, and I cant imagine any of these pre-TSO system were working without a User concept, since accountability (and Traceability etc) were a must.

I had the pleasure of moving an IMS security based User Management system, to RACF back in the mid-eighties, and this was a left over from a VM/CMS variant were customers IMS DB/DC system ran.



Hope it explains. And sorry I haven't had the time to pick source evidence for above reply.



Peter Rosenberg
Continue reading on narkive:
Loading...