In Rexx you can write routines that make all variables accessible to the called routine. You can also write routines that hide the caller's variables.
The following shows an example of a routine in which all variables are accessible:
/* Routine example */ i=10 /* Initialize I */ call myroutine /* Call routine */ say i /* Displays 22 */ exit /* End main program */ myroutine: /* Label */ i=i+12 /* Increment I */ return
The CALL instruction calls routine MYROUTINE. A label (note the colon) marks the start of the routine. A RETURN instruction ends the routine. Notice that an EXIT instruction is required in this case to end the main program. If EXIT is omitted, Rexx assumes that the following instructions are part of your main program and will execute those instructions. The SAY instruction displays 22 instead of 10 because the caller's variables are accessible to the routine.
You can return a result to the caller by placing an expression in the RETURN instruction, like this:
/* Routine with result */ i=10 /* Initialize I */ call myroutine /* Call routine */ say result /* Displays 22 */ exit /* End main program */ myroutine: /* Label */ return i+12 /* Increment I */
The returned result is available to the caller in the special variable RESULT, as previously shown. If your routine returns a result, you can call it as a function:
/* Routine with result called as function */ i=10 /* Initialize I */ say myroutine() /* Displays 22 */ exit /* End main program */ myroutine: /* Label */ return i+12 /* Increment I */
You can pass arguments to this sort of routine, but all variables are available to the routine anyway.
You can also write routines that separate the caller's variables from the routine's variables. This eliminates the risk of accidentally writing over a variable used by the caller or by some other unprotected routine. To get protection, use the PROCEDURE instruction, as follows:
/* Routine example using PROCEDURE instruction */ headcount=0 tailcount=0 /* Toss a coin 100 times, report results */ do i=1 to 100 call cointoss /* Flip the coin */ if result="HEADS" then headcount=headcount+1 /* Increment counters */ else tailcount=tailcount+1 /* Report results */ say "Toss is" result ||". Heads=" headcount "Tails=" tailcount end /* do */ exit /* End main program */ cointoss: procedure /* Use PROCEDURE to protect caller */ i=random(1,2) /* Pick a random number: 1 or 2 */ if i=1 then return "HEADS" /* Return English string */ return "TAILS"
In this example, the variable i is used in both the main program and the routine. When the PROCEDURE instruction is placed after the routine label, the routine's variables become local variables. They are isolated from all other variables in the program. Without the PROCEDURE instruction, the program would loop indefinitely. On each iteration the value of i would be reset to some value less than 100, which means the loop would never end. If a programming error causes your procedure to loop indefinitely, use the Ctrl+C key combination or close the command window to end the procedure.
To access variables outside the routine, add an EXPOSE operand to the PROCEDURE instruction. List the desired variables after the EXPOSE keyword:
/* Routine example using PROCEDURE instruction with EXPOSE operand */ headcount=0 tailcount=0 /* Toss a coin 100 times, report results */ do i=1 to 100 call cointoss /* Flip the coin */ say "Toss is" result ||". Heads=" headcount "Tails=" tailcount end /* do */ exit /* End main program */ cointoss: procedure expose headcount tailcount /* Expose the counter variables */ if random(1,2)=1 then do /* Pick a random number: 1 or 2 */ headcount=headcount+1 /* Bump counter... */ return "HEADS" /* ...and return English string */ end else tailcount=tailcount+1 return "TAILS"
To pass arguments to a routine, separate the arguments with commas:
call myroutine arg1, "literal arg", arg3 /* Call as subroutine */ myrc=myroutine(arg1, "literal arg", arg3) /* Call as function */
In the routine, use the USE ARG instruction to retrieve the argument.