Incremental read on DB table

[Former Member]

Hello,

I’m looking for a way to define the ‘

entry point

’ at the

incremental read

using select clause like:

Select * from {table} up to {m} rows
               where key-field1 = field1
               and     key-field2 = field2
               …
               and     key-fieldn = fieldn.

I will read rows, after next rows and so on.

How can I determine the correct where-condition to do realize an

incremental read

?

Thanks and best regards,

Kurt.

[Former Member]

Hi,

You are absolutely correct!! in this case you cannot use package size..

you can do one thing... take one of the main key fields, and fetch the data for a range of this key field at a time and process...

suppose that i want to process the data for 1 to 100 comany codes, i will process 10 company codes at a time..

Thanks and Best Regards,

Vikas Bittera.

[Former Member]

HI,

Why dont u use PACKAGE SIZE in the select query?

Thanks and Best Regards,

Vikas Bittera.

[Former Member]

Hello Vikas Bittera,

Thank you for your reply.

That’s clear; I can use <b>package size</b> as well as <b>up to n rows</b>, but I’m looking for the right way how to formulate the where-condition; if I have only one key, so I just can use <b>where key > key</b>. But if I have more key elements so I need a (complex) logical expression.

Do you have any idea?

Thanks and best regards,

Kurt

[Former Member]

Hi,

you need not to do so, as the records will come sorted by the primary key when u will use package size..

Thanks and Best Regards,

Vikas Bittera.

[Former Member]

Do you have any example?

Thanks & best regards,

Kurt.

[Former Member]

Hi,

SELECT vbeln erdat

FROM vbak

INTO TABLE li_vbak PACKAGE SIZE 50.

SELECT posnr matnr meins

FROM vbap

INTO TABLE li_vbap

FOR ALL ENTRIES IN li_vbak

WHERE vbeln = li_vbak-vbeln.

IF sy-subrc = 0.

• Now you have the two internal tables li_vbak and li_vbap filled

• with data.

• Do something with the data - remember to reinitialize internal

• tables

ENDIF.

ENDSELECT.

Thanks and Best Regards,

Vikas Bittera.

[Former Member]

Unfortunately I cannot use <b>select…endselect</b>, because I need a set of data to process IDOCs and posting data, I guess I cannot perform commit while select… endselect.

Is there another method to realize this?

Thanks and best regards,

Kurt

[Former Member]

Hi,

You are absolutely correct!! in this case you cannot use package size..

you can do one thing... take one of the main key fields, and fetch the data for a range of this key field at a time and process...

suppose that i want to process the data for 1 to 100 comany codes, i will process 10 company codes at a time..

Thanks and Best Regards,

Vikas Bittera.

[Former Member]

If you use Packet size, you do not need to specify number of rows or special keys - just give the full select and it returns a new packet of record each time it loops between the select and endselect statements.

Andrew

[RaymondGiuseppi]

You should define a CURSOR and then read the data package via this cursor.

* Declare cursor
DATA s_cursor TYPE cursor.
* Open Cursor (select)
      OPEN CURSOR WITH HOLD s_cursor FOR
        Select * from {table} 
               where key-field1 = field1
               and     key-field2 = field2
               …
               and     key-fieldn = fieldn.
* Then Fetch the data 
DO.
  FETCH NEXT CURSOR s_cursor
               INTO TABLE t_data
               PACKAGE SIZE package_size.
  IF SY-SUBRC <> 0. " No more data
    EXIT.
  ELSE.
    " Work with the data read
  ENDIF.
ENDDO.
* When you have read all the data, close the cursor
      CLOSE CURSOR s_cursor.

Regards