I forgot a third: the chads made great, if itchy confetti.
And a fourth: card correction by pushing chad(s) into holes before
punching new ones with a 12 key hand punch. This only worked if, like
us, you used optical card readers that didn't flex the cards.
There were some other advantages:
1. Content printing on each card. I could never, even when I was
handling paper tape a lot, read ASCII codes as fast as I could read
printed text.
I used to be able to read enough (newline, tab, space, numbers) to find
the right place on a tape.
As regards cards: our programmer's standby, the 12 key hand punch,
didn't print, so I learnt to read card codes at a good rate. Later we
were given printing hand punches but they were like a Dymo tape punch:
you had to dial the character and then hit to PUNCH bar to punch a
column. They were slow as hell: we hated them and used the old 12 key
punches by preference. I wish I'd had the sense to liberate one of the
12 key punches when they were phased out. They were marvelous Victorian
engineering: the best ones had cast iron bodies with a riveted-on brass
name plate saying "British Tabulating Machine Company". Their punches
never got blunt or jammed and they never wore out.
2. Ease of changes in the middle of a file. The two procedures for tape
were the one Roedy described above, and physical cut-and-splice.
Splicing increased the risk of mechanical problems.
>
Not if done right. I only used tape in anger at University to write
Algol 60 for an Elliott 503, the only machine I know that was faster at
floating point than integer arithmetic. Very appropriate seeing that it
was a scientific machine. But I digress....
We used to leave a foot or so of runout between procedure declarations
and in other suitable places, so we never had to copy & edit more than a
few feet of tape and splices never overlapped punched tape. IIRC we used
thin plastic heat-seal splicing tape. I don't remember having failed
splices or tape wrecks due to splices.
Contrast that with
inserting and removing cards in the middle of a card deck.
Actually, we only used a large program pack once and then slung them
because, even in 1968, we kept all program source on tape. Once a source
had been loaded we used small decks to edit the source on tape. The
programmer's overnight run started with a batch edit run that did
everybody's edits. This was followed by a batch compile. After that
individual test shots were run from the tape holding the compiled
programs. That was on an ICL 1900. By 1970 we'd moved our sources to
disk and the card decks had become individual edit/compile/test jobs for
George 1. A typical job pack would be no more than 50-100 cards. You
kept and reshuffled the commands, replacing the edits and test data as
needed.
I may be misremembering, but I have the impression that IBM mainframe
shops retained source as card decks a lot longer than we did. Certainly,
when I did a job in an IBM System/3 shop in NYC in 1976 all program
sources and, indeed, the master files as well were still on cards: those
nasty little 96 column jobbies.
Eee, lad. Tell that to the young people of today and they'll not believe
you.