Re: program defined Input file - examples please

["James W. Kilgore" <qappdsn@xxxxxxxxxxxxx>]

Dan,

It's been quite a while, but I don't think that you will have level
checks in RPGII.  So you can externally define your files and still use
the internal definition within RPGII.

If you are still running OCL, BLDINDEX will work as it always has.

The real gotchas for externally defining the files is where RPGII
programs do not properly initialize numeric fields when writing a
record.  I imagine that you could write a trigger program to handle this
situation once you have 100% confidence that you have a proper external
definition.

One of the things that we did when doing this sort of thing was in the
new name convention which made the transition a little easier.

For example:

In RPGII you would have file name DATAFILE and in the
input/calculations/output you would reference the file name for all
processing.

For the external definition you might create external file DATAEXT and
use DATAFILE as the record name.  Then after deleting file DATAFILE from
the system, create a logical called DATAFILE that shares the index with
DATAEXT.  This creates a file that can continue to be used with the
existing OCL and only costs about 4k of disk space due to the sharing of
access path.

It also preserves the name DATAFILE as the name used within the program.

Now, if you have multiple format flat files it does get a bit more
complex.  But still very doable.

hth,

Dan Bale wrote:
>
<<SNIP>>
 Just not sure what the caveats
> are, if any, for doing that.  Anybody have an RTFM that discusses potential
> gotchas for externalizing flatfiles?  How about alternate indexes?  The key
> would have to be that I wouldn't be required to recompile any program that
> uses a newly-converted-to-externally-described file.
>