Re: Externally-Described Field used as array name.

Mark wrote:
>  Will the ability to directly map a file format into a qualified DS be
one
>of the enhancements?  Until now "EDS" using the file in question would
>automatically populate the DS when a record is read.  Is there a way to
set
>a pointer to the file buffer instead of defining a superfluous DS?

I believe what we have in mind is something like this:  Today, you
can code a data structure name as the Result-Field operand on any
read/write I/O operation to a program-described file.  We will
likely just relax that restriction somewhat.


Njal replied to Mark:
>Why would you want to do a thing like that? And make your program
dependant
>on the physical implementation of how a physical file is stored.
>If the file had null indicators, variable length fields, date fields in
different
>formats, would you want to code for each possibility, and be prepared to
>change your code if the physical implementation changed?

First, the data buffer an RPG program sees is not necessarily how
the data is actually stored in the database.

Second, normally, you have to recompile programs anyways when the
record format changes, so there's no difference with receiving the
data buffer into a data structure rather than into distinct scalar
variables.  What we will likely require is that if a data structure
is specified as the Result-Field operand for an ext-desc I/O op,
the DS must itself be externally-described on the actual record
format.  (We're also thinking of some additional usability
enhancements in the area of ext-desc data structures.)

(Of course, nothing is official until you read that announce
letter.)

Cheers!  Hans

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