RE: *INPUTPACKED ignored for arrays ?

Timothy,

I think clean could only come from data conversion. Program described
files cannot help you keep your database clean, since the system cannot
perform any validation of your data. DDS described files are a little
more reliable, since field data types can be validated by the OS. The
only real problem with DDS data validation is that it happens when you
read it out of the file. It doesn't really keep you from stuffing bad
data into the file... This is where DDL described files take over. In
DDL described files, the DB blocks the write of obviously inappropriate
data.

As to the overlay issue, I never saw any code that seemed to define
any...

-Eric

-----Original Message-----
From: rpg400-l-bounces@xxxxxxxxxxxx
[mailto:rpg400-l-bounces@xxxxxxxxxxxx] On Behalf Of Timothy Adair
Sent: Tuesday, April 20, 2010 8:47 AM
To: rpg400-l@xxxxxxxxxxxx
Subject: Re: *INPUTPACKED ignored for arrays ?

Thanks for the reply. I did try this and got a clean compile but when I
ran
the pgm I got the dreaded decimal-data error. RPG doesn't allow you to
overly an array with an array, but I suppose I could double-define each
field (element). Surely there must be a cleaner way.




"Mark S. Waterbury" <mark.s.waterbury@xxxxxxx> wrote in message
news:mailman.23761.1271719215.2580.rpg400-l@xxxxxxxxxxxxxxx

Hello, Timothy:

Since it is a program-described file anyway, why not define the file
record layout as a series of packed fields, e.g.:

FLD1
FLD2
etc.

and define a data structure that allows you to overlay-define this

record

with the array? That way, you should be able to get the benefits of

the

RPG field validation on input, and still use the data as an array in

the

rest of the program.

Does that make sense?

All the best,

Mark S. Waterbury

Timothy Adair wrote:
I have an RPG IV pgm (not ILE) that is reading in an old QS36F
program-described file. This file has a "field" that is actually a
30-element packed numeric array (7,0). For records with valid data

in

this array, everything is fine; but I have some records with spaces

(hex

'4040404') in this array, and when it hits one of these, the program
gives a decimal-data error (understandably so).

I have specified the compile option of FIXNBR(*INPUTPACKED) but the
program is not fixing the data at input-time. Does *INPUTPACKED not

work

for input arrays? It seems to work for single fields.

And yes, I know, I need to fix the data. I intend to but that

doesn't

guarantee I won't encounter more of these in this file (or other

files).

I'd like to nail down a viable solution now. We will be converting

to

externally-described files eventually but for now this is the

situation I

have to deal with.

And yes, I have tried every conceivable combination of overlays &

arrays

to program around the problem. Every attempt either wouldn't

compile, or

gave the DD error when run.

Me duele la cabeza!

Ayuda?