On Wed, 6 Nov 2013 22:45:01 +0800, David Crayford wrote:
>JCL is neither simple or powerful. It's a piece of poorly designed junk that
>should never have made GA. Even it's original implementers admit that it's
>rubbish. Try explaining the reverse logic of condition codes to a youngster
>and they will die laughing.
>
>Hey, how do I do a loop in this code?
>
>Forget it kid, they didn't have rewind on punch card readers.
>
With all due respect (even I can muster all the respect due to JCL),
the declarative character, as opposed to procedural, of JCL provides
a couple advantages:
o It's possible to determine at initiation (some of) the resources
needed by a job and assure that they will be available, thereby
avoiding (some) deadlocks. (When we had our first couple
systems and no GRS, linkage editor reserve deadlocks were a
recurrent problem.)
o Auditors can tell by inspection who's doing what to whom. Thei
like that.
It would be impossible to do either in a Turing-complete command
language.
The "reverse logic of condition codes" probably was intuitive to an
assembler or FORTRAN programmer who thought of branching
around a statement.
DOS, like TSO, places the data set allocations before the command
("phase"?), which seems more natural than the JCL convention.
After all, which do you do first? Oops. I forgot: JCL is declarative
rather than procedural.
JCL met the needs of the 360 well. This is a different century.
-- gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
