It is also possible to defined new commands, these are called procedures. Procedures are like functions, but they do not return a value. Also, they generally should have side-effects (mutation or I/O).

This split between functions and procedures is explicit because it simplifies maintenance. A programmer does not need to think about side-effects when they are working with functions. Likewise, since procedures do not return values, there is no mixing of behavior.

Here is a simple example of a procedure:

  proc prompt (x:string) {
    write$ stdout, x; 
    fflush stdout;
  }
  
  prompt "Input string: ";
  println$ readln$ stdin;

Road Runner beats Coyote!

Input string: Road Runner beats Coyote!
Road Runner beats Coyote!

Here we notice:

• The proc binder introduces a procedure.
• The procedure prompt issues a prompt on stdout with a fflush to force the output to the device.
• Procedures have side effects, here it is reading and writing to I/O.
• Procedures cannot return values

Procedures can be used anonymously, just like functions. The syntax is practically the same.

  (proc (x:int) { println$ x; })

There's an even shorter version for situations when your procedure doesn't take any inputs:

  { println "Hello"; }

Since a proc cannot return a value, it can't be used in a map like fun functions do. The equivalent operator for procedures is called iter, which stands for iteration. With it, instead of mapping from one value to another, we visit each input and do some action.

Here is an example:

  var items = list(1,2,3);
  iter (proc (x:int) { println x; }) items;

1
2
3