Control Structures

if Expressions

The if special form allows us to evaluate one of two expressions based on a predicate. It takes in two required arguments and an optional third argument:

(if <predicate> <if-true> [if-false])

The first operand is what's known as a predicate expression in Scheme, an expression whose value is interpreted as either #t or #f.

The rules for evaluating an if special form expression are as follows:

  1. Evaluate <predicate>.
  2. If <predicate> evaluates to a truth-y value, evaluate and return the value if the expression <if-true>. Otherwise, evaluate and return the value of [if-false] if it is provided.

Can you see why this expression is a special form? Compare the rules between a regular call expression and an if expression. What is the difference?

Step 2 of evaluating call expressions requires evaluating all of the operands in order. However, an if expression will only evaluate two of its operands, the conditional expression and either <true-result> or <false-result>. Because we don't evaluate all the operands in an if expression, it is a special form.

Let's compare a Scheme if expression with a Python if statement:

Scheme Python 
scm> (if (> x 3)        
         1
         2)
 
>>> if x > 3:        
...     1
... else:
...     2

Although the code may look the same, what happens when each block of code is evaluated is actually very different. Specifically, the Scheme expression, given that it is an expression, evaluates to some value. However, the Python if statement simply directs the flow of the program.

Another difference between the two is that it's possible to add more lines of code into the suites of the Python if statement, while a Scheme if expression expects just a single expression for each of the true result and the false result.

One final difference is that in Scheme, you cannot write elif cases. If you want to have multiple cases using the if expression, you would need multiple branched if expressions:

Scheme Python 
scm> (if (< x 0)
         'negative
         (if (= x 0)
             'zero
             'positive
         )
 )
 
>>> if x < 0:
...     'negative'
... else:
...     if x == 0:
...         'zero'
...     else:
...         'positive'

cond Expressions

Using nested if expressions doesn't seem like a very practical way to take care of multiple cases. Instead, we can use the cond special form, a general conditional expression similar to a multi-clause if/elif/else conditional expression in Python. cond takes in an arbitrary number of arguments known as clauses. A clause is written as a list containing two expressions: (<p> <e>).

(cond
    (<p1> <e1>)
    (<p2> <e2>)
    ...
    (<pn> <en>)
    [(else <else-expression>)])

The first expression in each clause is a predicate. The second expression in the clause is the return expression corresponding to its predicate. The optional else clause has no predicate.

The rules of evaluation are as follows:

  1. Evaluate the predicates <p1>, <p2>, ..., <pn> in order until you reach one that evaluates to a truth-y value.
  2. If you reach a predicate that evaluates to a truth-y value, evaluate and return the corresponding expression in the clause.
  3. If none of the predicates are truth-y and there is an else clause, evaluate and return <else-expression>.

As you can see, cond is a special form because it does not evaluate its operands in their entirety; the predicates are evaluated separately from their corresponding return expression. In addition, the expression short circuits upon reaching the first predicate that evaluates to a truth-y value, leaving the remaining predicates unevaluated.

The following code is roughly equivalent (see the explanation in the if expression section):

Scheme Python 
scm> (cond
        ((> x 0) 'positive)
        ((< x 0) 'negative)
        (else 'zero))
 
>>> if x > 0:
...     'positive'
... elif x < 0:
...     'negative'
... else:
...     'zero'