Stan said:
>> The "crunch card" was a binary-punched card (you couldn't possibly
>> accidentally punch one on a keypunch).  It was also completely
> 
> Some computer I used, probably the Burroughs 6500,
> used a 1 & 2 overpunch as a control card.  (Punch a one, backspace,
> punch a 2 ... an illegal combination).
> Years later, IIRC, I discovered why: it caused the card reader
> to signal an exception to the main CPU.
> 
> Perhaps there was a similar underpinning to your crunch card?

No - it was home-made (I think created at UTK), and was an absolute job
delimiter. A little history follows.

I started at UTK in 1970 - the Admin side was a 360/40 DOS shop that had
replaced a 1401 installation. Student records, accounting, payroll and other
apps were run on it. Most accounting was 1401 mode, and new apps were
written mostly in Cobol. Within a year I was the systems department -
generated new supervisor images, created backup processes, assigned hard
drive space - all the little details required back then.

Academic Computing had a 360/65 running OS/MVT, HASP (spooler), and some
interactive stuff (TSO? ATS? I don't recall). Job names and such were highly
structured so student use could be reported to professors and research
groups could be charged correctly (actual monetary transactions were sent to
the 40 for processing). There were several places where students could walk
up and read their card decks into the system.

There was a lot of interest in publishing, using special print trains on the
1403 printers. There was a special carriage tape defined for the printer
(that still can't be easily emulated today), and some folks punched card
decks that could print pictures and other stuff. One problem was that you
couldn't trust that the previous user ended their deck right - your job card
deck could wind up as their data.

Plain vanilla HASP could be tricked, so the "crunch card" was born. As Jeff
noted, it was really full of holes - all legal EBCDIC punches - and it
didn't matter which way it went in. HASP required a job to begin with a
crunch card, and it was an absolute delimiter between jobs.

There was a second (secret) flavor of crunch card, used by the systems staff
only. When HASP found one of these, the subsequent job was allowed to do
special stuff - modify system libraries and such. These crunch cards were
highly controlled - systems team only.

In the 360 architecture, peripherals were generally responsible for validity
checking data. "Bad punched" cards would stop the reader, but the CPU would
only see it go "not ready" - no data was transferred. You -could- read
"column binary" on -some- readers, which was 2 bytes (16 bits) per column,
but that was seldom used.

And if I'm not mistaken, some codes could be punched several ways. The IBM
1620, for example, could read BCD numbers from (IBM?) time clocks. The digit
7 could be punched as 7, 12-7, 1-2-4, or 12-1-2-4. Even though a 1-2-4 punch
is not defined in most literature, I believe it could read in as a 7 on most
IBM 360 card readers of the time. Even though a general rule for EBCDIC is:
"any combination of 12/11/0/8/9 punches is valid but only one punch is
allowed in the 1-7 rows." Note that this only allows 224 codes.

At any rate, the "crunch card" Jeff speaks of was implemented purely as a
software mod in the version of HASP used at UT Knoxville. And it did the job
quite well :)

Back to the bushes - enjoy the holiday weekend, and thank a vet.
-- 
Regards  --  Tony B. Shepherd  --  [log in to unmask]

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *