[inforoots] CTSS & ITS

[John C Green Jr]

looks like inforoots has been alive long enough that we're
recycling topics.  My 26APR2004 comment on this is archived:

http://mail2.computerhistory.org/pipermail/inforoots/2004-April/000644.html

To further the discussion my foggy memory of a long ago conversation
with RLG ...

The modifications to the IBM 7094 included:

* Two banks of 32K Words, one for user programs, one for CTSS

* If an I/O instruction were executed in user memory the hardware
   would trap into CTSS

So what Compatible meant in CTSS was Compatible with stand alone.
There were no system calls for I/O.  You just executed the I/O
instruction, the hardware would trap, CTSS would figure out what
you were trying to do, and return control to you when it had
completed the I/O.

Considering how early CTSS was that was probably a good decision.
All existing programs designed to run on the raw hardware would
continue to work with no recoding.

What the Incompatible meant in ITS was NOT Incompatible with CTSS,
but Incompatible with stand alone PDP-6.  The ITS team thought it
was unreasonable to trap and emulate I/O.  It implemented system
calls.  And of course the names ITS vs. CTSS was an Project MAC
team joke.

Both of the PDP-6 architects, Gordon Bell & Alan Kotok, were
familiar with CTSS and the shortcomings of the 7094 for timesharing.
The PDP-6 was designed from the ground up for timesharing.  It was
the first system with vendor supplied and vendor supported
timesharing.  All previous timesharing systems were user written.

Lots of this is available:

http://www.catb.org/jargon/html/C/CTSS.html
http://www.catb.org/jargon/html/T/timesharing.html
http://en.wikipedia.org/wiki/PDP-6

Regards,
John Green