Log Function behaving diffently in SQL vs Pl/SQL

Question

Hello
This equation works OK in SQL but its implementation in PL/SQL throws an error:
SELECT CAST(LOG(2,32) as FLOAT) FROM DUAL;
Returns: 5
vs
x FLOAT := 0;
x := CAST(LOG(2,32) as FLOAT) ;
PLS-00382: expression is of wrong type
Incidentally I ran into a situation where LOG(2,32) return 5 in SQL but returns 4 where formatted TRUNC(LOG(2,32),0)
Any assistance would be appreciated

Answer 1

[email protected] wrote:
This is not so much specific to LOG function. More a so "rounding behavior" of SQLPLUS due to environment default settings.??? It has nothing to do with rounding. You'll get same PLS-00382 for pretty much any built-in function:
SQL> declare
  2  x FLOAT := 0;
  3  begin
  4  x := CAST(LOG(2,32) as FLOAT) ;
  5  end;
  6  /
x := CAST(LOG(2,32) as FLOAT) ;
ERROR at line 4:
ORA-06550: line 4, column 11:
PLS-00382: expression is of wrong type
ORA-06550: line 4, column 1:
PL/SQL: Statement ignored
SQL> declare
  2  x FLOAT := 0;
  3  begin
  4  x := CAST(POWER(2,32) as FLOAT) ;
  5  end;
  6  /
x := CAST(POWER(2,32) as FLOAT) ;
ERROR at line 4:
ORA-06550: line 4, column 11:
PLS-00382: expression is of wrong type
ORA-06550: line 4, column 1:
PL/SQL: Statement ignored
SQL> declare
  2  x FLOAT := 0;
  3  begin
  4  x := CAST(mod(3,2) as FLOAT) ;
  5  end;
  6  /
x := CAST(mod(3,2) as FLOAT) ;
ERROR at line 4:
ORA-06550: line 4, column 11:
PLS-00382: expression is of wrong type
ORA-06550: line 4, column 1:
PL/SQL: Statement ignored
SQL> Oracle still maintains two separate engines for SQL and PL/SQL. And the above bug is obviously related to that. Now same built-in function has separate implementation SQL and PL/SQL. In PL/SQL it is package STANDARD. At first I thought it is where bug manifests itself. However:
SQL> SELECT CAST(sys.standard.LOG(2,32) as FLOAT) FROM DUAL;
CAST(SYS.STANDARD.LOG(2,32)ASFLOAT)
                                  5
SQL> So most likely bug is caused by FLOAT implementation in PL/SQL.
SY.