Lab 8: Functions and Dynamic Scope

This is the archived website of SI 413 from the Fall 2012 semester. Feel free to browse around; you may also find more recent offerings at my teaching page.

This lab is due at 0900 next Tuesday, October 23. It should contain all the files in the starter code listed below, as well as a subfolder called tests with your tests. See the submit page for more info.

The starter code for this week's lab is... the solution to last week's lab. You can stick with your own solution or use mine (or some combination thereof). But in either case you should download the updated Makefile to keep things running smoothly.

Warm-up: Some more SPL coding!

Like last week, we are going to start by writing a small SPL program that, by the end of the lab, we should be able to correctly execute using our interpreter.

As an example of the kind of thing you need to do for this part, here is an SPL function that uses recursion to compute the value of n! - that is, n*(n-1)*(n-2)*...*1.

# Computes n!
new fact := lambda n {
  ifelse n <= 1
  { ret := 1; }
  { ret := n * fact@(n-1); }
# This should print 5! = 120
write fact @ 5;

Notice the structure of the lambda in SPL: we have the keyword lambda, followed by the name of the SINGLE argument (all functions are unary in SPL), followed by a block of code in curly braces for the body of the function. Remember also that ret is a special name that must be assigned to the return value.

The way functions are called is a little unusual: we use the FUNARG operator @, which you should know has highest precedence and is left associative. I assume you know what that means by now!


  1. Write SPL code for a recursive function called fib that takes a single argument n and computes the n'th Fibonacci number. So for example, after reading your code, typing write fib@10; should produce 55.
    Save this code in a file called ex1.spl, to be included with your submission.
    (No tests to submit for this exercise.)

Dynamic Scope with Central Reference Tables

In this part of the lab, we are going to extend the SymbolTable class to implement proper dynamic scoping rules. Recall from class that a CRT contains two parts: a mapping from strings to stacks of values (the bindings), and a stack of sets of strings (the variables in scope).

For this part of the lab, you just need to get dynamic scoping to work correctly with blocks and if/else/while statements in your SPL interpreter. You should do this by augmenting the SymbolTable class with some additional data structures, as well as two new methods, startScope() and endScope(), which do the obvious things. In what follows I will briefly outline one way of getting this working. But feel free to solve this in any reasonable way that makes sense to you.

It may be useful to reference the following C++ Standard Template Library APIs: stack, map, set.

Suggested Steps

Even if you don't follow all the steps like I suggest, I strongly recommend that you test your code often as you go along. The key to making a big change like this one successful is to do it in small, manageable steps, making sure everything works before you go on.


  1. Get dynamic scoping with basic control structures working. After this, a program such as
    { new x := 1;
      if x = 1 {
        new x := 2;
        write x;
      write x;
    should work and print 2 followed by 1.
    You are required to submit at least four tests for this exercise.

Lambdas and Function Calls

Now it's time to actually get function calls to work. This will require you to implement the eval() methods in the Lambda and Funcall classes. My suggestions are below, but remember that you are free to go your own way!



  1. Get function definitions and function calls to work. After this, you should be able to execute your ex1.spl program and compute Fibonacci numbers.
    You are required to submit at least six tests for this exercise.

Expression Statements

Right now, the following program should give an error:

new f := lambda x { write x + 5; };

What's going on? Well, the problem is that a function call like f(3) is an expression but not a statement. Now that we have function calls, we might want to have an expression as a statement all by itself, without having to write its result.

Your job in this part is to make this possible. This will involve adding a new subclass to the AST hierarchy - you might want to call it ExpStmt. It should be a subclass of Stmt that only has one part - an expression! Executing the statement will just mean evaluating the expression and throwing away the result.


  1. Allow semicolon-terminated expressions as single statements. This will require adding a new grammar rule to the spl.ypp Bison file as well as a new class to the ast.hpp file.
    You are required to submit at least two tests for this exercise.

OPTIONAL: More arguments, more returns

Important: Please save/submit your work before starting on this part. It is highly likely to break your code. I will be mightily impressed if you get everything here working, though. Keep in mind that you should be able to implement these changes with complete backward compatibility - that is, previously existing SPL programs should still work.


  1. (OPTIONAL) Allow multi-argument function definitions and function calls. You decide the exact syntax you would like for it, but you will probably need a new token to separate arguments, some new grammar rules, and significant changes to Lambda and Funcall. You might want to start with just two arguments instead of one.
  2. (OPTIONAL) Okay, I'm sure no one is going to do this. But just in case... Add the ability for functions to return multiple values. You can decide the exact syntax for this, but you might allow programs like:
    { new a := 0;
      new b := 0;
      new squareAndCube := lambda x {
        new sqr := x * x;
        ret := sqr, x*sqr;
      a, b := squareAndCube(3);
      write a; # Should print 9
      write b; # Should print 27