PL/SQL: Creating a Function: invalid identifier - plsql

Question

I have an oracle 19c ee database build via their docker image on Oracles github (https://github.com/oracle/docker-images/tree/master/OracleDatabase/SingleInstance). I am trying to follow their example as to how to create a function from here.
I have copied their example exactly. Setup table and data:
CREATE TABLE orders (
customer_id number(10),
order_total NUMBER(11,2)
);
INSERT INTO orders (customer_id, order_total) VALUES (1, 200.01)
The actual function:
CREATE FUNCTION get_bal(acc_no IN NUMBER)
RETURN NUMBER
IS acc_bal NUMBER(11,2);
BEGIN
SELECT order_total
INTO acc_bal
FROM orders
WHERE customer_id = acc_no;
RETURN(acc_bal);
END;
However I keep running into this error when I try and create the function
Query 2 ERROR: ORA-06550: line 5, column 27:
PL/SQL: ORA-00904: "ACC_NO": invalid identifier
ORA-06550: line 2, column 7:
PL/SQL: SQL Statement ignored
ORA-06550: line 6, column 7:
PLS-00372: In a procedure, RETURN statement cannot contain an expression
ORA-06550: line 6, column 7:
PL/SQL: Statement ignored
What am I doing wrong?

Answer 1

The example works for me. You must have mistyped something. Are you sure your function is exactly the same as the one in the manual?
ORA-00904: "ACC_NO": invalid identifier
suggests the declaration acc_bal NUMBER(11,2); is missing or different.
PLS-00372: In a procedure, RETURN statement cannot contain an expression
indicates that your code is a procedure, not a function.
SQL> CREATE TABLE orders (
2 customer_id number(10),
3 order_total NUMBER(11,2)
4 );
Table created
SQL> INSERT INTO orders (customer_id, order_total) VALUES (1, 200.01);
1 row inserted
SQL> CREATE FUNCTION get_bal(acc_no IN NUMBER)
2 RETURN NUMBER
3 IS acc_bal NUMBER(11,2);
4 BEGIN
5 SELECT order_total
6 INTO acc_bal
7 FROM orders
8 WHERE customer_id = acc_no;
9 RETURN(acc_bal);
10 END;
11 /
Function created
SQL> select get_bal(1) from dual;
GET_BAL(1)
----------
200.01
As an aside, while I'm a big fan of the Oracle documentation in general, and this example does neatly illustrate how to create a PL/SQL function, I think it could be improved:
For readability, it's better to give each declaration its own line, so line 3 would be better split into two with acc_bal NUMBER(11,2); on its own line.
The IS and AS keywords are interchangeable here, but surely create ... as (similar to what you might use to create a table or a view) reads better than create ... is.
Understandably, the author didn't want to complicate the example by introducing %type before it had been explained, but a more advanced version would use acc_bal orders.order_total%type; to make acc_bal inherit its datatype from the table column rather than hard-coding it. This goes for all three values used in the function.
The parameter and variable names are OK - they are at least clear - but there is a danger when using the same naming pattern for parameters and variables as for table columns. One day you will type WHERE c.customer_id = customer_id and wonder why you're getting more rows back than you expected. Again it's understandable that the author didn't want to get into that whole discussion in the first example, but it's something to think about. You might use get_bal.acc_no within the function, or use camelCase for parameters and variables, or prefix them with p_ for 'parameter' etc.
A basic rule of layout is that opening and closing keywords such as if/else and begin/end should be left-aligned. The END at line 10 is misaligned under its opening BEGIN. I suppose indenting the entire thing after the first line is a valid personal layout choice, but to me it doesn't add anything.
It's a good idea to leave blank lines around each SQL statement, to avoid a solid wall of text. Personally, I'd prefer a blank line before the RETURN at line 9.
A RETURN clause doesn't require any brackets. The compiler is ignoring the redundant brackets at line 9. I'd lose them.
It's good practice (though optional) to include the procedure/function name in the closing END, so line 10 would become END get_bal;
The COBOL-style uppercase habit is widespread in the industry, but there is no need for it. (PL/SQL's syntax is famously based on Ada, though some also point to ALGOL and PL/1 - nobody ever wrote those in uppercase.) I would improve readability by lowercasing the whole thing.
With these changes, I get this:
create or replace function get_bal
( inAccNo in orders.customer_id%type )
return orders.order_total%type
as
accBal orders.order_total%type;
begin
select order_total into accBal
from orders
where customer_id = inAccNo;
return accBal;
end;