The only slight inconsistency in what Rick Nuckolls suggests and how
Martin Phillips describes OpenQM, is that I see an implication that
OpenQM allows variables to have global scope if not explicitly
declared private.
We did this to preserve the general style of the language. Also, very often,
you actually want access to the variables in the main body of the program
without having to formally declare them but only want to make a few things
private to the subroutine.
I suggest we require IBM give us the ability to limit a variable's
(labels, too, so "symbol" may be a better word) scope to the local
subroutines.
We also have scope rules on label names in local subroutines such that it is
impossible to jump into or out of one as this would destroy the stacking of
the private variables.
Unfortunately, Pick style programmers have long known that you can jump
anywhere and leaping into or out of the middle of an internal subroutine is
so widespread that banning it would cause mass complaints. Of course, this
all comes down to an internal subroutine being nothing more than a labelled
statement in a monolithinc chunk of program.
Whether they are inherently limited to the private routines, or will
be global unless declared private, I am content to leave to IBM's
discretion. Maybe the OpenQM conformity would dictate
that. I think I'll go read their documentation.
The easiest way is to browse the online copy at www.openqm.docs/
Martin Phillips
Ladybridge Systems Ltd
17b Coldstream Lane, Hardingstone, Northampton, NN4 6DB
+44-(0)1604-709200
-------
u2-users mailing list
[email protected]
To unsubscribe please visit http://listserver.u2ug.org/
