15 Apr
2007
15 Apr
'07
2:35 a.m.
On 14 Apr 2007 at 19:57, woodelf wrote:
Chuck Guzis wrote:
Remembering APL di- and tri-graphs was a big struggle for me. Was
there ever a card punch code for APL?
There's something to be said for small
character sets.
No answer ... I guess not. Offhand I like the idea of
'<-'
instead of =. Also 5 or 6 character file names left much to be
improved on.
That's not a FORTRAN limitation! Remember that FORTRAN uses file
numbers, just like,uh--that other languge. Variables *were* limited
to 6 characters, but that restriction was lifted in later versions.
On the other extreme, I remember a gummint project involving about a
thousand GSA programmers and COBOL. Variable names were codified by
a rigidly enforced coding standard--something in the range of 30
characters long. Made for very tiresome reading of code, not to
mention punching. They really drove home the idea that "COBOL is
verbose".
I always liked the CDC SCOPE idea of permanent vs. temporary/local
files. You could have a permanent file with a very long name and any
number of generations. When you wanted to use it, you'd simply
"attach" it to a local file name that was accessible to your
programs. Other pernanent files weren't accessible to your programs
unless specifically "attached". Anything not specifically attached
to a permanent file or having a predefined disposition (e.g. OUTPUT,
PUNCH) was treated as a scratch file and discarded at job
termination.
The version idea was pretty neat--you normally would work with the
latest version of the file, but you could specify an earlier version
if desired.
Cheers,
Chuck