Card Columns (was Why did they make ... ?) 
Author Message
 Card Columns (was Why did they make ... ?)

Quote:

> Even on PL/I F, source lines of length up to 100 characters were
> allowed.  The default was SM=(2,72) but the maximum allowed was 100.  I
> always used SM=(2,100,1) for source files on disk and SM=(2,80,1) for
> source files on cards.

Allowed, yes.  But the norm was (2,72) -- or, in sane shops, (2,72,1).
(I added the column 1 convention, sequence checking on 73-80, and the
%INCLUDE statement to PL/I (D), myself.)

--
John W. Kennedy
"The poor have sometimes objected to being governed badly;
the rich have always objected to being governed at all."
   -- G. K. Chesterton, "The Man Who Was Thursday"



Wed, 06 Jul 2005 13:24:23 GMT  
 Card Columns (was Why did they make ... ?)
On Fri, 17 Jan 2003 09:51:19 -0700, "Russ Holsclaw"

Quote:

>A lot of folks on this newsgroup worked with minis, though, not
>mainframes, so they perceive the world a bit differently ... more
>"paper tape" than "card deck".

A lot of non-IBM mainframes worked with paper tape and
teleprinters long before IBM supported thousands of terminals
instead of RJE stations. I interacted with DEC-10, Univac 1100,
and GE machines before I even heard of IBM systems. JCL, what's
that? Why would anyone write a system as limited as that?

Thanks. Take care, Brian Inglis         Calgary, Alberta, Canada
--

    fake address                use address above to reply






Wed, 06 Jul 2005 16:46:23 GMT  
 Card Columns (was Why did they make ... ?)


(snip)

Quote:
> A lot of non-IBM mainframes worked with paper tape and
> teleprinters long before IBM supported thousands of terminals
> instead of RJE stations. I interacted with DEC-10, Univac 1100,
> and GE machines before I even heard of IBM systems. JCL, what's
> that? Why would anyone write a system as limited as that?

JCL is not user-friendly, but that doesn't make it limited.  It is
that it gives you so many options that make it hard to learn.
That, and that many don't have convenient default values.  The
ability to write different record formats, for example.  I would
say that systems that don't give you the options are limited.

-- glen



Wed, 06 Jul 2005 17:11:56 GMT  
 Card Columns (was Why did they make ... ?)


Quote:




>> If you want to have a discussion, DO NOT TOP POST, please.

Thank you.  That's much better.  Now I know the gist of the
conversation.
<snip>  Others are discussing this.

/BAH

Subtract a hundred and four for e-mail.



Wed, 06 Jul 2005 20:25:15 GMT  
 Card Columns (was Why did they make ... ?)


Quote:




>> If you want to have a discussion, DO NOT TOP POST, please.

>ok

>> Nope.  Go do some card sort work.  You also have to consider
>> how fast the operator is.  I preferred doing large decks because
>> I could stack and remove as the sorting happened.  I wouldn't say
>> double based on the card count.  It's the start and stop that
>> took the most time.

>As you say... the start and stop take pretty much the same time as long as
>your hoppers are big enough and assuming the operator
>can carry the cards...
>ie it takes about as long to start and stop a 100 card
>job as it does a 1000
>card job. The operator still has to make as many trips between the hoppers
>but just has a heavier load to carry.

What trips?  If count difference is 100 or 1000, there's one
trip.  If the count is higher, the operator dumped one bin at
a time into separate boxes.  I think I did this kind of job only
once or twice so I didn't enough practice to develop a better
technique.  There are others here who probably had a better
technique.

Quote:
> .. This would tend to make the algorithm
>slightly faster (per card) as the number of cards increases so its even
>better than O(n) !  If you stack and remove while it's happening its even
>better than that....Weirder and weirder :-)

If one is doing a large deck, the cards had to be removed while it
was happening.  It was sorta fun to figure out how to keep the
sorter fed _and_ not have to stop for evacuations.

/BAH

Subtract a hundred and four for e-mail.



Wed, 06 Jul 2005 20:29:57 GMT  
 Card Columns (was Why did they make ... ?)


Quote:
>On Fri, 17 Jan 2003 06:27:58 -0700, "Russ Holsclaw"



>>> Nevertheless they did it:

>>>  ... etc. etc.
>>> Actually, I don't think the first version of fortran even

>>> only way to introduce characters was in Holerith literals if I
>>recall correctly.

>>> You can encode anything if you're determined enough.

>>OK, I didn't argue that people didn't find a way to use TTY's to
>>adapt.
>>And, yes, I've been aware of the Figs and Ltrs mode on the model
>>15 since the very first day I saw one, about
>>45 years ago.

>>Nevertheless, I remain certain that the idea of compilers
>>scanning 1-72, and ignoring the last 8 columns of the
>>card originated with the row-binary card readers on those first
>>36-bit machines.

>>The alternate character set that PL/I suppported was to
>>accomodate the limited character sets of earlier keypunches and
>>printers, which could accomodate no more than 48 printer glyphs
>>(that's only 12 special characters beyond the letters and numeric
>>digits.)

>>Again... although it's an interesting coincidence that TTY's ALSO
>>had a 72-character line-length, I'm still holding to the
>>assertion that the practice of leaving out the last 8 columns in
>>the card was originally motivated by the limitations imposed by
>>the hardware in which FORTRAN was developed. Everything else
>>mentioned in your posts came later. Everything.

>Not all machines used punch cards -- many used paper tapes
>instead of cards! The tapes were prepped on teleprinters. See
>BAH's posts -- she started in DEC in TAPE PREP -- not keypunch.

Yup.  (I did card when I was working in college.)  

 <snip>

Once upon a time, there was somebody who submitted a job to
Tape Prep that involved creating a card deck.  I was the
only one in the group who could find the power on switch, let
alone program a drum card and operate the keyboard.  There
was one, and only one, keypunch in DEC.  It resided in payroll
(which was done on a Burroughs at the time...1971 or 1972).
The gals over there were very possessive about their machine
and it took them about five minutes of hovering over my head
to be reassured that I knew what I was doing.  Then it took
me another 1/2 hour to explain what I was doing.  IIRC, they'ld
never discovered how to use the auto-dup programming function,
nor the prog2 key.

So it took me 1/2 day to do a 20 minute job.

/BAH

Subtract a hundred and four for e-mail.



Wed, 06 Jul 2005 20:35:26 GMT  
 Card Columns (was Why did they make ... ?)


Quote:



>>     A programmer I once knew had a ready comeback for the
>> "Memory is cheap" pundits he'd encounter in design sessions:
>> He'd hold out his hand, palm up, and invite them them to fill
>> it with memory ...

>Have you noticed how much RAM you can hold in the palm of your
>hand
>these days? More than any computer had when the in-place sort
>algorithms were invented!

<snip>

You're missing a very important point; sort is a dual operation.
Without a lot of fussing, you don't compare three things at a time.

Size of memory means nothing other than an immediate access to
the next item to be sorted.

/BAH

Subtract a hundred and four for e-mail.



Wed, 06 Jul 2005 20:39:52 GMT  
 Card Columns (was Why did they make ... ?)


<snip>

Quote:
>A lot of folks on this newsgroup worked with minis, though, not
>mainframes, so they perceive the world a bit differently ... more
>"paper tape" than "card deck".

I suggest that you examine your assumption w.r.t. the newsgroup
alt.folklore.computers.

/BAH

Subtract a hundred and four for e-mail.



Wed, 06 Jul 2005 20:44:14 GMT  
 Card Columns (was Why did they make ... ?)

Quote:

> > Even on PL/I F, source lines of length up to 100 characters were
> > allowed.  The default was SM=(2,72) but the maximum allowed was 100.  I
> > always used SM=(2,100,1) for source files on disk and SM=(2,80,1) for
> > source files on cards.

> Allowed, yes.  But the norm was (2,72) -- or, in sane shops, (2,72,1).

We used (1, 80), to avoid the inevitable excursion beyond column
72.
(a problem that was ever-present with FORTRAN.)
Quote:
> --
> John W. Kennedy



Wed, 06 Jul 2005 21:17:17 GMT  
 Card Columns (was Why did they make ... ?)
On Sat, 18 Jan 2003 09:11:56 GMT, "glen herrmannsfeldt"

Quote:



>(snip)
>> A lot of non-IBM mainframes worked with paper tape and
>> teleprinters long before IBM supported thousands of terminals
>> instead of RJE stations. I interacted with DEC-10, Univac 1100,
>> and GE machines before I even heard of IBM systems. JCL, what's
>> that? Why would anyone write a system as limited as that?

>JCL is not user-friendly, but that doesn't make it limited.  It is
>that it gives you so many options that make it hard to learn.
>That, and that many don't have convenient default values.  The
>ability to write different record formats, for example.  I would
>say that systems that don't give you the options are limited.

IMO systems where you don't need JCL are better designed for
humans to use.  I never had a problem picking up and using JCL
myself, but most application programmers did not have two clues
about what numbers to plug in for dataset allocation, because
they did not (want to) know device track sizes and record
overhead counts, numbers of tracks and cylinders.

Thanks. Take care, Brian Inglis         Calgary, Alberta, Canada
--

    fake address                use address above to reply






Wed, 06 Jul 2005 22:18:29 GMT  
 Card Columns (was Why did they make ... ?)

Quote:



> If you want to have a discussion, DO NOT TOP POST, please.

>>Well I got a lot of replies very fast from that but to take some of
>>you to task...

>>The card sort IS O(n) because of the simple fact that doubling the
>>number
> of
>>cards EXACTLY doubles the time taken to sort those cards.

> Nope.  Go do some card sort work.  You also have to consider
> how fast the operator is.  I preferred doing large decks because
> I could stack and remove as the sorting happened.  I wouldn't say
> double based on the card count.  It's the start and stop that
> took the most time.

Okay, I confess that I learned everything I know about sorting from the
relatively small number of Prof. Corbat's introductory lectures that I
somehow managed not to sleep through some thirty-odd years ago (no
reflection on Corby; I just wasn't very keen on getting out of bed in
the morning when I was a freshman).

BUT -- wouldn't that still be O(n)?  I mean, we generally speak in
terms of an arbitrarily large n, don't we?  The O(0) components of the
operator's time starting and stopping the sorter and switching columns
and so on are {*filter*} when n is small, but it gradually disappear in
the noise (perhaps literally, if there's enough card equipment around!)
as n becomes large.  In other words: to any particular desired degree
of accuracy, doubling the card count will double the sort time if the
card count is sufficiently high.  (That's doesn't quite amount to
"EXACTLY" but it's as near as we get to exact in the real world!)

OTOH, my practical experience with card sorters was pretty much
confined to dealing with my own dropped program decks, so I am prepared
to be corrected.

--
Roland Hutchinson              Will play viola da gamba for food.


remove spam.  If your message looks like spam I may not see it.



Wed, 06 Jul 2005 23:49:07 GMT  
 Card Columns (was Why did they make ... ?)

Quote:

>The default for IBM System/360 PL/I was to use columns 2 - 72 for
>the source code but IIRC that could be overridden with a compiler
>option.
>System/360 used the string "/*" in columns 1 & 2 of an input card
>deck to indicate end of file; thus it would not be possible to
>start a comment in column 1.  IIRC, it was possible to substitute
>a different EOF indicator through some option on the "//SYSIN DD *"
>JCL statement, but I never learned it well enough to remember.

Using the DLM option was the only way to read a job stream as data.

//ddname  DD  *,DLM=xx

...which for environments where you have a user-operated card reader
(and especially so in a university environment) is an invitation for
hacking (bad definition).  Someone would leave read in a job stream
with a DD *, DLM=xx card followed by a waste card.  If the EOF button
hadn't been pressed, the reader would read and process the DD * card,
then stop.  The hacker now uses NPRO to remove the waste card and
walks away; the next user's entire deck is now the input data stream
to the hacker's job.

At my PPOE we solved this problem by requiring via local mods to HASP
that every JOB card be immediately preceeded by a locally defined card
(named for obscure reasons the "CRUNCH" card).  This card was what
I called an "absolute delimiter" which could never be read as
data, so (assuming that the second user in the above example had put
a CRUNCH card at the top of his deck), the first user's job would be
closed with an empty input stream, and the second user's job would
be processed as it should be.

Users were "encouraged" to get into the habit of using CRUNCH cards
by code in HASP that flushed any JOB card that it recognized unless
the immediately previous card was a CRUNCH card.

Quote:
>Also, I knew (and still know) how to set up a drum card for an 029
>keypunch (a skill that doesn't seem to be much in demand these
>days :-)).

But it lets you understand the war stories we tell about drum cards!

Joe Morris



Thu, 07 Jul 2005 01:07:51 GMT  
 Card Columns (was Why did they make ... ?)

Quote:


>> A lot of non-IBM mainframes worked with paper tape and
>> teleprinters long before IBM supported thousands of terminals
>> instead of RJE stations. I interacted with DEC-10, Univac 1100,
>> and GE machines before I even heard of IBM systems. JCL, what's
>> that? Why would anyone write a system as limited as that?
>JCL is not user-friendly, but that doesn't make it limited.  It is
>that it gives you so many options that make it hard to learn.
>That, and that many don't have convenient default values.  The
>ability to write different record formats, for example.  I would
>say that systems that don't give you the options are limited.

...and JCL is designed to support batch operations, which is what
the mainframe world used in general-purpose data centers when the
S/360 line and its associated operating systems were being designed.

Admittedly, it took IBM several tries before its grafting of a
time-sharing interface onto OS/360 was decent.  (Did anyone actually
*like* TCAM?)  That's one of the reasons why VM had such success:
its primary environment was remote interactive computing, and as was
noted a in some postings a few days ago, CMSBATCH was really no
competition for OS/360 and its successors if you wanted to run
production batch work.

Joe Morris



Thu, 07 Jul 2005 01:16:45 GMT  
 Card Columns (was Why did they make ... ?)

Quote:

> On Fri, 17 Jan 2003 18:55:11 GMT, "John W. Kennedy"


>>>On Fri, 17 Jan 2003 04:40:00 GMT, "John W. Kennedy"





>>>>>>>>Default seems to be at col 72, in my experience.  Can't

>>>>>>imagine why....

>>>>>>>>:->  I mean, who's going to add a comment in the final 8

>>>>>>cols??

>>>>>>>Columns 73-80 were for card numbering.

>>>>>>I've mentioned this before, but for those who missed it:

>>>>>>The tradition of using only 1-72 in mainframe compilers dates
>>>>>>back to the 704 through 7094 computers, whose online card
>>>>>>readers only read in row-binary, and would read two 36-bit
>>>>>>words from each row of punches, 24 words per card.
>>>>>>Conversion to characters was done by software.

>>>>>>Columns 73-80 were ignored because the reader literally
>>>>>>couldn't pick them up, unless you dropped in
>>>>>>another control panel that was jumpered differently.
>>>>>>[...]

>>>>>Moreover some ttys could print only 72 columns.

>>>>Yes, but TTY's were not used on the IBM 704.  The fact that FORTRAN,
>>>>JCL, COBOL, PL/I -- virtually everything used on IBM mainframe computers
>>>>except RPG -- ignores columns 73-80 depends on that bit of 50's hardware.

>>>There were lots of other hardware available in the 50s that used
>>>paper tapes and TTYs.

>>Yes, but there was very little use of paper tape and/or TTY's on IBM
>>mainframes -- and IBM's own TTY-like devices, in the 60's, were
>>wide-carriage Selectrics.

> Not disputing that, but I don't think IBM was a {*filter*} computer
> company globally until after the 360 line came out, and prior to
> that there were a lot of other mainframe companies doing things
> differently, with different rules for their versions of languages
> e.g. fixed column input was one possible input format on systems
> that might not have card readers.

There were a good many US manufacturers -- Burroughs, Sperry-Rand,
National Cash Register, Control Data, Honeywell, RCA, General Electric
(collectively known as the "seven dwarfs") -- even Philco made a stab at
it at one point, and had a FORTRAN/ALGOL hybrid language called ALTAC.
But IBM held a commanding presence in the US business world almost from
the first, partly because they already dominated the pre-computer
punched-card-accounting business.  (IBM's machines had been entirely
programmable by the customer; Remington-Rand's machines had frequently
needed an engineer or even a trip to the factory.)

When the 360 came out, RCA simply copied its problem-state architecture
(though they had an incompatible supervisor state, requiring their own
operating systems), and Honeywell's entire advertising campaign, for
years, was on the theme, "We can move you from an IBM 1401 to a
Honeywell system more easily than IBM can move you to a 360."

--
John W. Kennedy
"The poor have sometimes objected to being governed badly;
the rich have always objected to being governed at all."
   -- G. K. Chesterton, "The Man Who Was Thursday"



Thu, 07 Jul 2005 04:44:30 GMT  
 Card Columns (was Why did they make ... ?)

Quote:


>> Note that RPG was born on the 1401, a machine on which the COBOL
>> compiler could take _hours_ to run.  RPG was the Turbo Pascal of its
>> era.  It is also a very easy language to code, within its particular
>> problem domain (which constituted a very sizable fraction of
>> dollars-and-cents programming in those days).
> Are you sure?  I started with IBM in June of 1960 in Cincinnati and I
> don't recall seing a 1401 until sometime in 1961.  While I didn't use it
> myself, I recall hearing about RPG from people who I thought were
> working on 650's and 704's and/or 705's at the time.

That's a little before my time (my first program was written in 1965 in
Basic FORTRAN on Brown's 7070) but there are 1401 hardware artifacts in
the RPG language, and the MIT Press "History of Early IBM Computers"
says quite explicitly that RPG was developed for the 1401, and first
delivered "early in 1961", "about three months" after the first 1401
delivery.  And, strategically, the 1401 needed RPG as no prior machine had.

--
John W. Kennedy
"The poor have sometimes objected to being governed badly;
the rich have always objected to being governed at all."
   -- G. K. Chesterton, "The Man Who Was Thursday"



Thu, 07 Jul 2005 04:44:35 GMT  
 
 [ 373 post ]  Go to page: [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25]

 Relevant Pages 
 

 
Powered by phpBB® Forum Software