The Select Query

Question

Hi All,
  SELECT matnr FROM mara INTO TABLE gt_mara
                               WHERE matnr IN s_matnr
                               AND   mtart = p_mtart.
  IF NOT gt_mara[] IS INITIAL.
    SORT gt_mara BY matnr.
    SELECT aufnr objnr aufpl FROM caufv INTO TABLE gt_caufv
                             FOR ALL ENTRIES IN gt_mara
                             WHERE aufnr IN s_aufnr
                             AND auart = p_auart
                             AND werks = p_werks
                             AND plnbez = gt_mara-matnr.
  ENDIF.
I have replaced the above for all entries query by the below inner join. Found in the debugg both have same number of records in the (DEV server). Performance wise also this one is better. Let me know both are exactly same or not.
  SELECT a~matnr
         b~aufnr
         b~objnr
         b~aufpl
         INTO TABLE gt_caufv FROM mara AS a INNER JOIN caufv AS b
         ON amatnr = bplnbez WHERE matnr IN s_matnr
                                     AND mtart =<b> p_mtart</b>
                                     AND aufnr IN s_aufnr
                                     AND auart = p_auart
                                     AND werks = p_werks.
My concern with the join is <b>amatnr = bplnbez</b>  is whether this will give me unique records or duplicate records. It seems to be good and unique in the DEV but not sure about the PROD server .  Kindly Clarify.
Thanks,

Answer 1

The join by itself will allow duplicate records if they exist, since the same material could exist in more than one order.
However, the FOR ALL ENTRIES addition de-dups the result set; this is a feature of FOR ALL ENTRIES.   This forces the original to be distinct apart from any other consideration.
If you want to guarantee the records are unique, why not just add DISTINCT?
e.g.
      SELECT DISTINCT
             a~matnr
             b~aufnr
             b~objnr
             b~aufpl
             INTO TABLE gt_caufv
             FROM mara AS a
             INNER JOIN caufv AS b
             ON a~matnr = b~plnbez
             WHERE matnr IN s_matnr
               AND mtart = p_mtart
               AND aufnr IN s_aufnr
               AND auart = p_auart
               AND werks = p_werks.
Alternatively, you could SORT the resulting itab and DELETE ADJACENT DUPLICATES.
In the case of DISTINCT, the sorting and deleting of duplicates occurs on the database server; in the case of explicit SORT and DELETE ADJACENT DUPLICATES, it occurs on the application server.   This may be a consideration for performance.
Lastly, the join and FOR ALL ENTRIES are different in one other way.
If the itab gt_mara in FOR ALL ENTRIES is empty, the WHERE clause is ignored and the whole table is loaded.   This is a feature of FOR ALL ENTRIES, which you can confirm in SAP HELP.   If this is not the behavior that you expect, you need to check for it.
Good luck.
Brian
Edited for spelling, wording
Message was edited by:
        Brian Sammond