Re: Externally-Described Field used as array name.

Peter wrote:
>Hi Patrick,
>
>I'll try<G>:
>
>The compiler sees PDLB02 in the data structure first, and allocates memory
>for it there.
>
>Then it sees PDLB02 in the input specs and wants to allocate memory for it
>within the input buffer, but it can't have a unique field name pointing to
>two different memory locations.

No, coding an input field allocates storage for the variable only
if there is no other definition of the variable.  You can easily
specify a particular variable as an input field and also define it
with a D-Spec.  The D-Spec definition specifies explicitly how the
variable is the be stored internally within the program.  If there
is no explicit definition of the variable, the information on the
Input-Field spec is used to implicitly define the variable.  For
example, numeric variables implicitly get defined as packed
(regardless of the external attributes of the numeric field).

>
>I believe with V5R1 you are able to use the data structure name (which
your
>example doesn't have) as a qualifier to the field name, so you could
indeed
>have the duplicate name.  But you'd have to be aware that the value in one
>would not automatically be copied to the other.

Indeed!  This is perhaps the most important enhancement to RPG in
V5R1, and will likely be expanded on in future releases.  Code
keyword QUALIFIED on the DS spec, and subfields within the data
structure are referred to using the DSNAME.SUBFNAME syntax.  This
is an important enhancement - so important that I believe that
programming standards in RPG shops should be updated to require
all data structures to be qualified if possible.

Cheers!  Hans

Hans Boldt, ILE RPG Development, IBM Toronto Lab, boldt@ca.ibm.com