Returning values

A function can return values with the RETURN instruction.

Definining the returned types in a function declaration

Function definitions can specify the list of data types returned by the function with the RETURNS clause.

Note: The RETURNS clause in the function header is not mandatory. However, it is recommended, to define the complete function signature, and take advantage of better code checking by the compiler.

If the function returns a single value, specify the data type after the RETURNS clause:

FUNCTION get_count() RETURNS INT
  ...
END FUNCTION

Note: When the function returns a single value/type, it is also possible to enclose the type in parentheses:

FUNCTION get_count() RETURNS (INT)

But for convenience, the parentheses are optional when only one return type is used.

If the function returns a several values, you must specify the data types after the RETURNS clause inside parentheses:

FUNCTION get_address() RETURNS ( CHAR(5), VARCHAR(100), INTEGER )
  ...
END FUNCTION

The RETURNS types specification can use base data types such as INTEGER, user-defined types, and classes.

SCHEMA stores
TYPE t_cust DYNAMIC ARRAY OF RECORD LIKE customer.*
...
FUNCTION get_cust_list() RETURNS t_cust
  ...
END FUNCTION

Returning from the function

Use the RETURN instruction in the body of the function, to push a list of values on the stack, and return to the caller.

The RETURN instruction takes an optional comma-separated list of expressions:

FUNCTION get_address() RETURNS ( CHAR(5), VARCHAR(100), INTEGER )
  DEFINE zipcode CHAR(5),
         street VARCHAR(100),
         city INTEGER
  ...
  RETURN zipcode, street, city
END FUNCTION

The next example shows a function returning a single value:

FUNCTION get_count()
   ...
   RETURN count
END FUNCTION

If a function does not need to return a value, the RETURN instruction can be used without arguments:

FUNCTION show_notfound()
   IF SQLCA.SQLCODE==0 THEN
      RETURN
   END IF
   ...
END FUNCTION

Using functions in expressions

Functions returning a simple value can be used in expressions:

LET c = get_count()

Returning a list of values

Funtions returning multiple values must be invoked with a CALL instruction using the RETURNING clause. Values specified in RETURN statement must correspond in number and position to the RETURNING clause of the CALL instruction, and must be of the same or of compatible data types, to the variables in the RETURNING clause of the CALL statement. An error results if the list of returned values in the RETURN statement conflicts in number or in data type with the RETURNING clause of the CALL statement that invokes the function.

MAIN
  DEFINE zipcode CHAR(5),
         street VARCHAR(100),
         city VARCHAR(50)
  CALL get_address() RETURNING zipcode, street, city
END MAIN

FUNCTION get_default_address()
   RETURN "00000", "<undefined>", "<undefined>"
END FUNCTION

Returning complex structures

When returning simple built-in types like INTEGER, values are copied on stack and copied to the caller variables.

When returning a RECORD structure, all values are expanded on the stack:

TYPE t_rec RECORD
         pkey INT,
         name VARCHAR(20)
     END RECORD

MAIN
    DEFINE r1 t_rec
    CALL get_rec() RETURNING r1.*
    DISPLAY r1.*
END MAIN

FUNCTION get_rec()
    DEFINE r t_rec
    LET r.pkey = 999
    LET r.name = "Mike"
    RETURN r.*
END FUNCTION

When returning complex types such as objects or dynamic arrays, the reference of the element are copied on the stack (this means that you can create an object inside a function, and return its reference in the RETURN statement):

MAIN
    DEFINE c base.Channel
    LET c = open_file("myfile.txt")
    ...
END MAIN

FUNCTION open_file(filename)
    DEFINE filename STRING
    DEFINE c base.Channel
    TRY
        LET c = base.Channel.create()
        CALL c.openFile(filename, "r")
        RETURN c
    CATCH
        RETURN NULL
    END TRY
END FUNCTION