IBM Mainframe Forum

Hi,

I can't understand, how DL1 handles the key feedback area.

I do a GU on a root segment with a key, that is not in the database.
The key contains more than one field and I only know the first one, the rest I put to null.

The ASMTDLI returns a segment level 00 and a status code GE, but a KFB lenth of 0.
Nevertheless, if I do a GN, DLI knowns the current position and returns the next root.

Why do I not get the key in the KFB?

I hope sombody can explain this to me!

Mike

Your position never changed during the bad call. That does NOT mean the position didn't exist.

You were positioned SOMEWHERE before the bad call, right? So if you skip the bad call, you could still do the GN and get a valid segment.

The first unqualified GN on a given PCB will return the first segment in the database, the same as if you start with an unqualified GU.

Try adding a few unqualified GNs before your bad GU call, and see if makes more sense to you.

I think you need to look at generic keys - the first call using key (say key*) positions the location immediately before the first segment with a key beginning 'key'. It does not position ON asegment but BETWEEN so it does not retrieve anything. The GN will then retrieve the record that is currently immediately after this location.

Ed Goodman wrote:Your position never changed during the bad call. That does NOT mean the position didn't exist.

You were positioned SOMEWHERE before the bad call, right? So if you skip the bad call, you could still do the GN and get a valid segment.

The first unqualified GN on a given PCB will return the first segment in the database, the same as if you start with an unqualified GU.

Try adding a few unqualified GNs before your bad GU call, and see if makes more sense to you.

Sorry, but that is not the behavior I got (and described).

The GU is the first call and if a bad call does not change the position, why do I get with a GN the next root after the used key?
I defititly do not get the first root!

NicC wrote:I think you need to look at generic keys - the first call using key (say key*) positions the location immediately before the first segment with a key beginning 'key'. It does not position ON asegment but BETWEEN so it does not retrieve anything. The GN will then retrieve the record that is currently immediately after this location.

Yes, that is exactly the situation!
My question is: Why do I not get the key* in the key feedback area?

the key of which - th GU or the GN? You do not get a key in the KFB for a GU because you are not positioned ON a segment but between segments. If it is for the GN then I do not know - you would have to check in the manual to see what it says.

I mean the GU!

DL1 stored the position before a root, because the following GN gets the next root.
But it seams, that DL1 stores the position only in an internal KFB, not in the PCB.

Is this the normal behavior of DL1?
Can I do something, to get the KFB in the PCB? (without changing the call)

As no segment has been retrieved there is no key to save. IMS probably has a cyl/track/sector reference so that it knows where to advance from. Whatever, until it retrieves a segment there is nothing to feed back except the success, or otherwise, of the call. I believe it is in the manual. If it is not in the manual than it doesn't happen.

The KFB contains everything that IMS was able to successfully get. In your case, that was nothing, so the KFB length is 0.

I've lost track a little bit... Does your GU SSA have a key 'EQ' or a key 'GE'. ? That changes the behavior somewhat.

The GU SSA have a EQ ( = ) and it gets a GE as status code.

If I summarize all the answers correct, the normal behavior is, that DL1
- clears the KFB after a not successful call
- fills the KFB only after a successful call
- always keeps the position internally independed of the success