On this page:
let
let*
letrec
let-values
let*-values
letrec-values
let-syntax
letrec-syntax
let-syntaxes
letrec-syntaxes
letrec-syntaxes+ values

2.9 Local Binding: let, let*, letrec, ...

+Local Binding in The Racket Guide introduces local binding.

(let ([id val-expr] ...) body ...+)
(let proc-id ([id init-expr] ...) body ...+)
The first form evaluates the val-exprs left-to-right, creates a new location for each id, and places the values into the locations. It then evaluates the bodys, in which the ids are bound. The last body expression is in tail position with respect to the let form. The ids must be distinct according to bound-identifier=?.

Examples:

> (let ([x 5]) x)

5

> (let ([x 5])
    (let ([x 2]
          [y x])
      (list y x)))

'(5 2)

The second form evaluates the init-exprs; the resulting values become arguments in an application of a procedure (lambda (id ...) body ...+), where proc-id is bound within the bodys to the procedure itself.

Example:

> (let fac ([n 10])
    (if (zero? n)
        1
        (* n (fac (sub1 n)))))

3628800

(let* ([id val-expr] ...) body ...+)
Similar to let, but evaluates the val-exprs one by one, creating a location for each id as soon as the value is available. The ids are bound in the remaining val-exprs as well as the bodys, and the ids need not be distinct; later bindings shadow earlier bindings.

Example:

> (let* ([x 1]
         [y (+ x 1)])
    (list y x))

'(2 1)

(letrec ([id val-expr] ...) body ...+)
Similar to let, but the locations for all ids are created first and filled with #<undefined>, and all ids are bound in all val-exprs as well as the bodys. The ids must be distinct according to bound-identifier=?.

Example:

> (letrec ([is-even? (lambda (n)
                       (or (zero? n)
                           (is-odd? (sub1 n))))]
           [is-odd? (lambda (n)
                      (and (not (zero? n))
                           (is-even? (sub1 n))))])
    (is-odd? 11))

#t

(let-values ([(id ...) val-expr] ...) body ...+)
Like let, except that each val-expr must produce as many values as corresponding ids, otherwise the exn:fail:contract exception is raised. A separate location is created for each id, all of which are bound in the bodys.

Example:

> (let-values ([(x y) (quotient/remainder 10 3)])
    (list y x))

'(1 3)

(let*-values ([(id ...) val-expr] ...) body ...+)
Like let*, except that each val-expr must produce as many values as corresponding ids. A separate location is created for each id, all of which are bound in the later val-exprs and in the bodys.

Example:

> (let*-values ([(x y) (quotient/remainder 10 3)]
                [(z) (list y x)])
    z)

'(1 3)

(letrec-values ([(id ...) val-expr] ...) body ...+)
Like letrec, except that each val-expr must produce as many values as corresponding ids. A separate location is created for each id, all of which are initialized to #<undefined> and bound in all val-exprs and in the bodys.

Example:

> (letrec-values ([(is-even? is-odd?)
                   (values
                     (lambda (n)
                       (or (zero? n)
                           (is-odd? (sub1 n))))
                     (lambda (n)
                       (or (= n 1)
                           (is-even? (sub1 n)))))])
    (is-odd? 11))

#t

(let-syntax ([id trans-expr] ...) body ...+)

Creates a transformer binding (see Transformer Bindings) of each id with the value of trans-expr, which is an expression at phase level 1 relative to the surrounding context. (See Identifiers and Binding for information on phase levels.)

The evaluation of each trans-expr is parameterized to set current-namespace to a namespace that shares bindings and variables with the namespace being used to expand the let-syntax form, except that its base phase is one greater.

Each id is bound in the bodys, and not in other trans-exprs.

(letrec-syntax ([id trans-expr] ...) body ...+)

Like let-syntax, except that each id is also bound within all trans-exprs.

(let-syntaxes ([(id ...) trans-expr] ...) body ...+)

Like let-syntax, but each trans-expr must produce as many values as corresponding ids, each of which is bound to the corresponding value.

(letrec-syntaxes ([(id ...) trans-expr] ...) body ...+)

Like let-syntax, except that each id is also bound within all trans-exprs.

(letrec-syntaxes+values ([(trans-id ...) trans-expr] ...)
                        ([(val-id ...) val-expr] ...)
   body ...+)
Combines letrec-syntaxes with a variant of letrec-values: each trans-id and val-id is bound in all trans-exprs and val-exprs.

The letrec-syntaxes+values form is the core form for local compile-time bindings, since forms like letrec-syntax and internal-definition contexts expand to it. In a fully expanded expression (see Fully Expanded Programs), the trans-id bindings are discarded and the form reduces to a combination of letrec-values or let-values, but letrec-syntaxes+values can appear in the result of local-expand with an empty stop list.

For variables bound by letrec-syntaxes+values, the location-creation rules differ slightly from letrec-values. The [(val-id ...) val-expr] binding clauses are partitioned into minimal sets of clauses that satisfy the following rule: if a clause has a val-id binding that is referenced (in a full expansion) by the val-expr of an earlier clause, the two clauses and all in between are in the same set. If a set consists of a single clause whose val-expr does not refer to any of the clause’s val-ids, then locations for the val-ids are created after the val-expr is evaluated. Otherwise, locations for all val-ids in a set are created just before the first val-expr in the set is evaluated.

The end result of the location-creation rules is that scoping and evaluation order are the same as for letrec-values, but the compiler has more freedom to optimize away location creation. The rules also correspond to a nesting of let-values and letrec-values, which is how letrec-syntaxes+values for a fully-expanded expression.

See also local, which supports local bindings with define, define-syntax, and more.