IBM MAINFRAME HELP & SUPPORT FORUMS

sudhakar_mainframe · Joined: 29 Jun 2006 Posts: 7

HI,

I have the following logic to initialize the Large arrays in Cobol.

Declaration in WS.
---------------------
05 WS-TEMP.

07 A OCCURS 100 TIMES

Procedure division statements.
-----------------------------------
INITIALIZE A (1)

MOVE WS-TEMP TO WS-TEMP (LENGTH OF A (1) + 1 : )

How this logic increases the performance?...

Thanks,
Sudhakar.

Bill O'Boyle · Joined: 14 Jan 2008 Posts: 345 Location: Orlando, FL, USA

What is the picture clause of array "A" or its elementary items (if any)?

Bill

CICS Guy · Joined: 18 Jul 2007 Posts: 1199 Location: At my desk

An old Assembler trick....(IIRC)
MVI X'40',WS-TEMP
MVC WS-TEMP+1(L'WS-TEMP-1),WS-TEMP
Overlapping move to propagate the first byte to the rest of the field.

Bill O'Boyle · Joined: 14 Jan 2008 Posts: 345 Location: Orlando, FL, USA

Guy,

I stopped using initialization-method this once COBOL2 was released. Although, it did work well in OS/VS COBOL.

But in this example, if the array is to be cleared to SPACES, then MOVE SPACES to the 05 level will generate (most likely) an MVCL and hence, 5 instructions to perform this initialization or a series of MVC's (as you've shown).

Also, since COBOL2, a VALUE clause can be used for each elementary item of an array and then the entire array can be cleared in the program Prologue code.

Bill