RE: Multi-field dynamic array

 >> that under MicroFocus COBOL for the PC that I did prove that the system
did physically expand the array when I defined it as varying between x and
y.

In some older versions of COBOL (mostly PC) this probably did happen - as
you note -t hey were concerned about memory footprint.  I think MF may have
changed it later for reasons of compatibility with mainframe COBOL.

 >> In the regression and recursion routines I've done it sure 'acts' like
it does truly expand, but that could be just as you said - it's pretending
to.

If you are using the ILE COBOL support, then yes it does expand.  But that's
not the same thing.  The reason it expands is because each invocation level
creates its own set of local variables.

I'm quite willing to be proven wrong - the ILE compiler may have other
differences from the OPM one, but for a variety of reasons I doubt this is
one of them.  If you think about it, if you do not pre-allocate the full
array size, then it may have to "move" around in memory since a contiguous
extension block may not be available.  That would invalidate any pointers to
the array itself or any elements with no way of detecting it had happened.
I suspect the COBOL compiler writers would have wanted to avoid that.


Jon Paris
Partner400

www.Partner400.com
www.RPGWorld.com