🚀 go-pugleaf

In article <877@cbmvax.cbmvax.cbm.UUCP>, grr@cbmvax.cbm.UUCP (George Robbins)
writes:
>                       ... Now was TOPS-10 a new products, or just a marketing
> name for the older monitor program, along with the revisionistic DECsystem-10
> name?

Well... a little of both. The "Monitor" (as it was known in those days) had
stabilized (?!?!) at version 4S72 (major version 4, "S"wapping support, minor
rev 72), and was getting patched to death with bug fixes. (I think the last
4-series we ran at Emory Univ. was 4S72AJ, where "AJ" means the 26+10th patch;
they got up to 4S72BN or something finally... yeah, 52+14 patches!.) At that
point there was a *massive* rewrite/enhancement/extension, and the major
version bumped to "5". At the same time, "Monitor" became "TOPS-10", and
"PDP-10" became "DECsystem-10".

We're talking about an entirely new TTY driver ["line discipline" to Unix
folk], file system (and disk format... worse than the 4.1 ==> 4.2 move),
and new batch system, spoolers, etc., so maybe the name change wasn't all
that bogus (...but the old hands still called it "the monitor".) It was like
jumping from Unix v.6 to Berkeley 4bsd.

The new system eventually became rock-solid, but the first releases were
murder (worse than 4.2 with no bug fixes!). The 5.01 release wouldn't
stay up very long; 5.02C was sort of o.k. for a handful of users, if you
didn't try to run the batch system or spoolers ("you did WHAT!?!? (*crash*)
(*tinkle*) (*tinkle*)"); 5.02D would survive running with the line printer
spooler active; and by 5.03C the system would take a pretty good beating
from batch & users and keep on going. As I recall, 5.04 was flakey again,
but 5.04A fixed that. 6.01 added paging (in addition to swapping), but had
so many "improvements" it took 'til 6.02A to settle down. (Also, 6.02A was
the last version that was supported in the non-paging KA-10 processor.)

Incidently, even old 4Snn ("10/50 monitor") had the "JACCT" bit ("setuid root")
in the late 1960's, but remember that unlike Unix, the list of JACCT programs
was compiled into the kernel! (And because TOPS-10 was a "job"-oriented system
rather than "process"-structured, the kernel had to really worry about making
sure the JACCT bit got turned off no matter how a JACCT program died, to keep
the user from inheriting the priviledge.) Let's see, the JACCT programs were
LOGIN, LOGOUT, SYSTAT ("ps"), QUEUE ("lpr/lpd"), DAEMON ("init"), and so on.


Rob Warnock
Consultant
{amdcad,sun,fortune}!redwood!rpw3