Looping Issue with AT NEW / END

[Former Member]

Hi All,

I'm having some problem with some abap codes, its looping an ITAB with control break of "AT NEW f1" and "AT END f1", the problem is that even thought the value in "f1" field is identical for sy-tabix 1 and 2, but somehow when it hits AT END f1 its still invoking the event even though its just the first loop of the index, and after the first loop it consider the 2nd index as a new value when it hits AT NEW event. The funny part is that this happens in QAS landscape but not in DEV, so its working fine in DEV, so was wondering any guru did encounter this problem before. below is an illustration of the codes. btw the OS is running on LINUX SUSSE, so not sure whether to dwell into that level.

The ITAB[] value are something like this - note INDEX denotes the sy-tabix index value

INDEX	XBLNR
1	00034567
2	00034567

LOOP AT itab.

     AT NEW xblnr.

          PERFORM subroutine_1.

     ENDAT.

     ........some processes.

     AT END OF xblnr.

          PERFORM subroutine_2.

     ENDAT.

ENDLOOP.

what i was expecting was that the 1st loop the AT NEW event should be invoke, then then would skip AT END event. During the 2nd loop, it would skip AT NEW event and invoke AT END event.

What actually happens is that the 1st loop it invokes AT NEW (should happen) and also AT END(-this shouldnt happen) event, during the 2nd loop it invokes AT NEW (this shouldnt happen) and AT END (should happen)event

I can assure you that during the debug session, the XBLNR value are the same for the header line and also the ITAB[] lines. i tested this in DEV, its just working fine, but once in QAS its just puzzles me. anyone encounter this b4 pls advise.

Regards,

Shimano

[SuhaSaha]

Hello Shimano,

Is XBLNR the left-most column of your internal table? If no, then the behavior of AT NEW & AT END are as expected.

From SAP documentation on control break statements:

 The control breaks take place when the content of the component compi or another component to the left of compi changes.

BR,

Suhas

[rosenberg_eitan]

Hi,

I was wondering what is the logic behind:

"The control breaks take place when the content of the component compi or another component to the left of compi changes."

Regards.

[Former Member]

It means you need to define the fields in the internal table in the order that the field  values changes from left to till the field that you specify in AT NEW or AT END OF. The remaining fields are irrelevant for your purposes and will have values **** as its contents inside these loops.

If you are interested in only one field changes you have to declare that field as the first field in the internal table mentioning that other fields are irrelevant for looking for changes in their contents.

This is only in case of one field value change. But ABAP also allows you the flexibility of observing multiple field changes based on your requirement. In this case you have to declare the fields from left to right indicating that when the first field  value changes you are not interested in the remaining field contents. When the first field value remains same then only you are interested in a change in the second field value and so on till the field you specified at AT NEW or AT END OF.

Hope its clear.

[rosenberg_eitan]

Hi,

Thanks for you explanation.

So if we want to generate different report based on differet sorting for each report we need to define different structure ?

If I am right it seem a little bit restricting syntax and not very productive....

Regards.

[Former Member]

Yes in this kind of situation you have to go for it in that way only. But this kind of situation rarely occurs as normally you will always have key fields in the beginning of the structure and will look for a change in that.

[rosenberg_eitan]

Hi,

Rarely occurs ? I wander.

Take for example classic case:

Customer,Product,Date,Value,Quantity.

It is very common requirement to summarize the same data in different sorting and different control field.

With ALV you can define sorting and control field dynamically. I believe that this flexibility should be also

available in ABAP .

I can think on different technics to overcome this limitation but this will create unnecessary complexity in the code.

Regards.

[Former Member]

In such cases, you already know what a programmer does, write your own code handling multiple field changes

[Former Member]

Hi Shimano,

If you use AT NEW & AT END statement then XBLNR should be first field in your internal table itab..in the same sorting order

Most prob this will solve ur issue

Thanks

Ben

[Former Member]

Hi Shimano,

Your program is not at all working strangely. Suhas is right.

Please first understand how AT NEW and AT END OF works in the below link.

http://help.sap.com/saphelp_nw2004s/helpdata/en/9f/db9f1f35c111d1829f0000e829fbfe/content.htm

then you can solve your problem yourself...

thanks.

[Former Member]

.Hi Shimano,

At new , at first will check the field value from the leftmost column.

In this scenario at new xblnr and at end xblnr will be triggered twice because the value of index is changing.

If either of index or xblnr will change the statements will be triggered.

Put your Xblnr field as first column of the internal table.

Regards,

Swarna

[jrg_wulf]

If you're not interested in using "SUM" as a control break result, i don't see why you would want to use AT NEW / END in the first place.

You might want to try ON CHANGE OF field_name instead, since it is invoked on (and only on) changes of field_name . If you need to do some postprocessing at the end of a group of data, you could use a save_workarea for the previous entry.

Regards

Jörg

[Former Member]

Hi Jorg,

Please press F1 on ON CHANGE OF statement and see. Please don't suggest obsolete statements.

thanks.

[SuhaSaha]

Hello Jörg,

Afaik ON CHANGE OF is an obsolete syntax

BR,

Suhas

[Former Member]

Hi All,

Appreciate the help, was reading through the Control Breaks documentation stating that The control breaks take place when the content of the component compi or another component to the left of compi changes. Apparently there are some other fields to the left of compi - XBLNR, thus the invoke of the event, though if XBLNR is to be used for the control break criterion, XBLNR would best to be furthest left field to be defined in the ITAB. Below is an illustration of the ITAB in both landscape QAS and DEV

QAS - there were change of result to the left of XBLNR that caused it to recognise as an END/ NEW scenario.

INDEX	BLART	XBLNR
1	SA	00034567
2		00034567

DEV - identical field value prior to XBLNR

INDEX	BLART	XBLNR
1	SA	00034567
2	SA	00034567

To anyone out there who are puzzled with the similar case study, the missing BLART at the 2nd index in QAS was causing the problem.

[venkatapavan_sinagam]

Index field Should be  char c

Regards,

Sinagam.