RE: Level Check Issue

Martin,

Wasn't that a "bug" though :-)

Crispin.

-----Original Message-----
From: rpg400-l-bounces@xxxxxxxxxxxx [mailto:rpg400-l-bounces@xxxxxxxxxxxx]
On Behalf Of Martin Rowe
Sent: Tuesday, January 15, 2008 6:53 PM
To: RPG programming on the AS400 / iSeries
Subject: Re: Level Check Issue

On 15/01/2008, Crispin <cbates@xxxxxxxxxxx> wrote:

Jim,

Just to be clear. You can compile a file from the same DDS as many times

as

you want, and on as many machines as you want. The format level id will
always be the same (given the same CCSID - and I'm not sure that CCSID is
actually a problem as far as Format Level Id is concerned, but _may_ cause

a

problem with CPYF, just know I ran into an issue with CCSID's recently,

but

could be mistaken on that point).

It shouldn't matter about the order that the files/programs are created.

If

the DDS _is_ the same, you will get the same format level id...

...Except when IBM introduce a fix for the way that format level
identifier is calculated in certain circumstances, and you end up with
a restored file that has a different format-level ID to the saved
version. We hit this a few years back at V5R1 or V5R2 after some PTFs
were put on and we started to get level checks. Our programs were
compiled on the dev system, over files replicated by HA software.
When we put them onto the production system they level-checked.

Turned out the PTFs had introduced a bug with some SQL generated files
that contained date or timestamp fields. You could do a CRTDUPOBJ and
the level ID would stay the same. Do a SAVOBJ-RSTOBJ and it was
different. Once we identified it was behaviour introduced by the PTFs
IBM got a fix to us quickly, but it was certainly confusing for a
while.

And if anyone wants to automate the check for potential level checks
in advance, there's my open source CHKLVLCHK utility -
http://www.dbg400.net/cgi-bin/twiki/view/DBG400/ChkLvlChk

Regards, Martin