Hi Matthew,
thank you for the support regarding my initial post.
Well, I tried to get the answer from the documentation.
My question is about a table declared as SORTED TABLE WITH UNIQUE KEY table_line.
What about a LOOP ... WHERE <field> = <value>?
For the database, we know that an index is used, when then first field is (all fields are) first part of the primary index (or another index). Using index field n without giving index field n-1 results in sequential read.
In my internal table, this would be the field MANDT because it is declared TYPE <ddic table>. For database, implicitly SY-MANDT will be used. To be sure I specified mandt = sy-mandt in the WHERE clause and the the following but not all fields.
The first n fields (say field1 ..fieldn) I know and I can specify as <field> = <value>. The last fields (field n+1 ... field n+x) are the attributes I am interested in. They are also part of the key table_line.
The only thing I found is about a table of references of the table structure using table_line-><component> for the access. That will probably result in code some of my colleagues would not fully understand.
So my question is not really answered. I could not find information about using the key named table_line. Will the compiler know that my fields used in the WHERE clause are part of the key table_line and optimize accordingly?
Actually it would cost me some time to setup a measure routine to check. So I still hope for some experts guidance. Of course, everyone who feels he can add more to the solution than posting links it invited to help. Raymond: Thank you for the links. I coudn't find the string 'table_line', please help.
Thanks and regards
Clemens