Primitive Expressions

Just like in Python, atomic, or primitive, expressions in Scheme take a single step to evaluate. These include numbers, booleans, symbols.

scm> 1234    ; integer
1234
scm> 123.4   ; real number
123.4

Symbols

Out of these, the symbol type is the only one we didn't encounter in Python. A symbol acts a lot like a Python name, but not exactly. Specifically, a symbol in Scheme is also a type of value. On the other hand, in Python, names only serve as expressions; a Python expression can never evaluate to a name.

scm> quotient      ; A name bound to a built-in procedure
#[quotient]
scm> 'quotient     ; An expression that evaluates to a symbol
quotient
scm> 'hello-world!
hello-world!

Booleans

In Scheme, all values except the special boolean value #f are interpreted as true values (unlike Python, where there are some false-y values like 0). Our particular version of the Scheme interpreter allows you to write True and False in place of #t and #f. This is not standard.

scm> #t
#t
scm> #f
#f

Call Expressions

Like Python, the operator in a Scheme call expression comes before all the operands. Unlike Python, the operator is included within the parentheses and the operands are separated by spaces rather than with commas. However, evaluation of a Scheme call expression follows the exact same rules as in Python:

1. Evaluate the operator. It should evaluate to a procedure.
2. Evaluate the operands, left to right.
3. Apply the procedure to the evaluated operands.

Here are some examples using built-in procedures:

scm> (+ 1 2)
3
scm> (- 10 (/ 6 2))
7
scm> (modulo 35 4)
3
scm> (even? (quotient 45 2))
#t

Special Forms

The operator of a special form expression is a special form. What makes a special form "special" is that they do not follow the three rules of evaluation stated in the previous section. Instead, each special form follows its own special rules for execution, such as short-circuiting before evaluating all the operands.

Some examples of special forms that we'll study today are the if, cond, define, and lambda forms. Read their corresponding sections below to find out what their rules of evaluation are!

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?

scm> (if (> x 3)
         1
         2)

>>> if x > 3:
...     1
... else:
...     2

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):

scm> (cond
        ((> x 0) 'positive)
        ((< x 0) 'negative)
        (else 'zero))

>>> if x > 0:
...     'positive'
... elif x < 0:
...     'negative'
... else:
...     'zero'

The special form define is used to define variables and functions in Scheme. There are two versions of the define special form. To define variables, we use the define form with the following syntax:

(define <name> <expression>)

The rules to evaluate this expression are

1. Evaluate the <expression>.
2. Bind its value to the <name> in the current frame.
3. Return <name>.

The second version of define is used to define procedures:

(define (<name> <param1> <param2> ...) <body> )

To evaluate this expression:

1. Create a lambda procedure with the given parameters and <body>.
2. Bind the procedure to the <name> in the current frame.
3. Return <name>.

The following two expressions are equivalent:

scm> (define foo (lambda (x y) (+ x y)))
foo
scm> (define (foo x y) (+ x y))
foo

define is a special form because its operands are not evaluated at all! For example, <body> is not evaluated when a procedure is defined, but rather when it is called. <name> and the parameter names are all names that should not be evaluated when executing this define expression.

All Scheme procedures are lambda procedures. To create a lambda procedure, we can use the lambda special form:

(lambda (<param1> <param2> ...) <body>)

This expression will create and return a function with the given parameters and body, but it will not alter the current environment. This is very similar to a lambda expression in Python!

scm> (lambda (x y) (+ x y))        ; Returns a lambda function, but doesn't assign it to a name
(lambda (x y) (+ x y))
scm> ((lambda (x y) (+ x y)) 3 4)  ; Create and call a lambda function in one line
7

A procedure may take in any number of parameters. The <body> may contain multiple expressions. There is not an equivalent version of a Python return statement in Scheme. The function will simply return the value of the last expression in the body.