First column's data is missing

Question

I am a newbie to ABAP. I am trying this program with open sql, and when I execute the program, the first column's data is always missing. I have looked up and the syntax appears to be correct. I am using kna1 table, the query is pretty simple too. If anybody notice the issue, please help me out.

DATA: WA_TAB_KNA1 TYPE KNA1,
      IT_TAB_KNA1 TYPE TABLE OF KNA1,
      V_KUNNR TYPE KUNNR.

SELECT-OPTIONS: P_KUNNR FOR V_KUNNR.

SELECT name1 kunnr name2
       INTO TABLE IT_TAB_KNA1 FROM KNA1
       WHERE KUNNR IN P_KUNNR.

LOOP AT IT_TAB_KNA1 INTO WA_TAB_KNA1.
  WRITE:/ WA_TAB_KNA1-KUNNR,' ', WA_TAB_KNA1-NAME1.
ENDLOOP.

Answer 1

This is a classic - I suppose every ABAP developer has to experience this at least once.

You're using an internal table of structure KNA1, which means that your target variable has the following structure

 ccckkkkkkkkkklllnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNN...

with ccc being the client, kkkkkkkkkk being the field KUNNR (10 characters), lll the field LAND1 (3 characters), then 35 ns for the field NAME1, 35 Ns for the field NAME2 and so on.

In your SELECT statement, you tell the system to retrieve the columns NAME1, KUNNR and NAME2 - in that order! This will yield a result set that has the following structure, using the nomenclature above:

nnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnkkkkkkkkkkNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNN

Instead of raising some kind of type error, the system will then try to squeeze the data into the target structure - mainly for historical reasons. Because the first fields are all character fields, it will succeed. The result: the field MANDT of your internal table contains the first three characters of NAME1, the field KUNNR contains the characters 4-13 of the source field NAME1 and so on.

Fortunately the solution is easy: use INTO CORRESPONDING FIELDS OF TABLE instead of INTO TABLE. This will cause the system to use a fieldname-based mapping when filling the target table. As tomdemuyt mentioned it's also possible to roll your own target structure -- and for large data sets, that's a really good idea because you're wasting a lot of memory otherwise. Still, sometimes this is not an option, so you really have to know this error - recognize it as soon as you see it and know what to do.