[TUHS] Lions laboratory and a short excursion into DOS-11

[Clem Cole via TUHS]

On Sat, Aug 15, 2026 at 6:04 PM Luther Johnson via TUHS <tuhs at tuhs.org>
wrote:

> I think I read a book by C. A. R. Hoare in the mid-70's or early 80's
> that used the term "monitor" as something like a monolithic kernel,
> where, once a thread entered it, it was not re-scheduled and nothing
> could happen behind its back, the monitor was a critical region, and
> only one thread could be in at a time - therefore whatever it did inside
> was guaranteed to finish, all the data it managed was left in a coherent
> state on exit, and so, on the next entry by any other thread. Similar
> but a little different than the idea of an ground-level interactive
> program, I have seen the term used in both ways, since.
>
Yes - Hoare used that term for a *synchronization primitive*, which, like
Dijkstra's P/V semaphores (from the THE paper BTW) allows concurrent
processes to operate.
https://en.wikipedia.org/wiki/Monitor_(synchronization)     IIRC, he
recommended that they be built into the compiler and a number of parallel
languages, such Tunis' "Concurrent Euclid."

But Hoare's monitors should not be confused with the term used to describe
an entire operating system/executive.


> On 08/15/2026 02:36 PM, Clem Cole via TUHS wrote:
> > below
> >
> > On Sun, Aug 9, 2026 at 6:11 AM Lars Brinkhoff via TUHS <tuhs at tuhs.org>
> > wrote:
> >
> >> Aron Insinga wrote:
> >>> Calling the operating system the Monitor is, I think, from the
> >>> PDP-6/PDP-10 Monitor (later renamed TOPS-10)
> >> There's also BBN's TENEX monitor (ObTuhs: tcsh is the "TENEX csh"), of
> >> which there there is a resident part and a swappable (demand paged)
> part.
> >>
> >> I wager there were many other operating systems - also outside DEC -
> >> around that time using monitor in a similar sense.
> >>
> > Lots of systems used that term.  As John pointed out, the IBM 709x IBSYS
> > manuals describe the System Monitor; IBJOB has the Program Monitor,
> *etc*.
> > Aron and Lars point out that TOPS, Tenex, and WAITS all use that term,
> > which I suspect they inherited from the earlier PDP-6 Monitor.   IIRC,
> even
> > the ITS folks called it a Monitor.  But I think that was a historical
> > artifact, and by the time many of us started working as OS developers in
> > the early/mid 1970s, the world had moved on; although some, like Doug,
> who
> > lived the transitions, probably have a better understanding.
> >
> > As I understand the history, the term "Monitor" stems from the idea of a
> > program that "monitors" or supervises user execution. In early systems
> like
> > DEC's RT-11, you can see the explicit breakdown: ISTR, the user-interface
> > command-line component, was called the Keyboard Monitor (KMON), while the
> > resident (in-core) portion that remained in memory to route I/O
> interrupts
> > was called the RMON (Resident Monitor). A key point is that this was for
> > OS-like systems before MMUs were on all systems and timesharing, or at
> > least multitasking, was considered "de rigueur."   By the mid-to-late
> > 1970s, as architectural theories had all matured to require protected
> > memory spaces and a greater level of hardware abstraction (which was
> > obviously accelerated by the realization of security models that
> > timesharing required), such that the term "Kernel" emerged to describe
> the
> > innermost, privileged core of software stripped of its outer user-facing
> view
> > of the system (which also begets the term "shells").
> >
> > I also suspect a piece of this was that, in early systems, the command
> > interpreter was built into the core "executive."   But that started to
> > change after Dijkstra presented at the inaugural ACM Symposium on
> Operating
> > Systems Principles and published it in 1968 in the *CACM *(Volume 11,
> Issue
> > 5, pages 341–346), his paper "The Structure of the 'THE'-Multiprogramming
> > system" described the system being built in layers and in particular the
> > "console layer" as being separate from the "processor allocation" and
> > "memory segment control."  By the mid-to-late 1970s, except on the
> smallest
> > microprocessors (*i.e.* CP/M and later MS DOS), some amount of Dijkstra's
> > layering and modularity in the OS was pretty much standard; thus the term
> > "monitor" fell out of use.
> >
> > Of course, as OS developers, many of us later took Dijkstra's idea
> further
> > and developed the concept of a microkernel and a fully layered OS (which
> I
> > still think are better than "monolithic ones", even if they don't eke out
> > the absolute maximum from the hardware because of their overhead).
> >
> > Clem
>
>