Re: [open-cobol-list] Problem with pointers

[Keisuke Nishida]

David Korn wrote:

> The redefine allows the pointer to be treated as an integer but
> because of byteswapping the two do not have the same representation.
> Using COMP-5 for the redefines instead of COMP would fix this problem.
> However, this seems to happen hundreds of times throughout the 3 million
> lines of Cobol I am trying to port, so it would be nice if the
> compiler automatically treated a redefines for a pointer to be COMP-5
> rathern than COMP. 
>
> Can this be changed?  If not, can you let me know where the change
> would have to be made?

Hmm, you could do it by the following patch:

--- cobc/field.c.~1.12.~        2004-03-08 06:47:37.000000000 +0900
+++ cobc/field.c        2004-07-25 21:46:52.000000000 +0900
@@ -298,6 +298,10 @@
   /* validate REDEFINES */
   if (f->redefines)
     {
+      if (f->redefines->usage == CB_USAGE_POINTER
+         && f->usage == CB_USAGE_BINARY)
+       f->usage = CB_USAGE_COMP_5;
+
       /* check OCCURS */
       if (f->redefines->flag_occurs)
 	cb_warning_x (x, _("the original definition `%s' should not have 
OCCURS"),

However, I don't think it is a good idea to merge this patch
into open-cobol.

Keisuke