On this page:
defproc
defproc*
defform
defform*
defform/ subs
defform*/ subs
defform/ none
defidform
defidform/ inline
specform
specsubform
specsubform/ subs
specspecsubform
specspecsubform/ subs
defparam
defboolparam
defthing
defstruct*
defstruct
deftogether
racketgrammar
racketgrammar*
defidentifier
schemegrammar
schemegrammar*
current-display-width
4.3.3 Documenting Forms, Functions, Structure Types, and Values

syntax

(defproc maybe-kind maybe-id prototype
         result-contract-expr-datum
         pre-flow ...)
 
prototype = (id arg-spec ...)
  | (prototype arg-spec ...)
     
arg-spec = (arg-id contract-expr-datum)
  | (arg-id contract-expr-datum default-expr)
  | (keyword arg-id contract-expr-datum)
  | (keyword arg-id contract-expr-datum default-expr)
  | ellipses
  | ellipses+
     
maybe-kind = 
  | #:kind kind-string-expr
     
maybe-id = 
  | #:id [src-id dest-id-expr]
     
ellipses = ...
     
ellipses+ = ...+
Produces a sequence of flow elements (encapsulated in a splice) to document a procedure named id. Nesting prototypes corresponds to a curried function, as in define. The id is indexed, and it also registered so that racket-typeset uses of the identifier (with the same for-label binding) are hyperlinked to this documentation.

A defmodule or declare-exporting form (or one of the variants) in an enclosing section determines the id binding that is being defined. The id should also have a for-label binding (as introduced by (require (for-label ....))) that matches the definition binding; otherwise, the defined id will not typeset correctly within the definition.

Each arg-spec must have one of the following forms:

(arg-id contract-expr-datum)

An argument whose contract is specified by contract-expr-datum which is typeset via racketblock0.

(arg-id contract-expr-datum default-expr)

Like the previous case, but with a default value. All arguments with a default value must be grouped together, but they can be in the middle of required arguments.

(keyword arg-id contract-expr-datum)

Like the first case, but for a keyword-based argument.

(keyword arg-id contract-expr-datum default-expr)

Like the previous case, but with a default value.

...

Any number of the preceding argument. This form is normally used at the end, but keyword-based arguments can sensibly appear afterward. See also the documentation for append for a use of ... before the last argument.

...+

One or more of the preceding argument (normally at the end, like ...).

The result-contract-expr-datum is typeset via racketblock0, and it represents a contract on the procedure’s result.

The decoded pre-flow documents the procedure. In this description, references to arg-ids using racket, racketblock, etc. are typeset as procedure arguments.

The typesetting of all information before the pre-flows ignores the source layout, except that the local formatting is preserved for contracts and default-values expressions. The information is formatted to fit (if possible) in the number of characters specified by the current-display-width parameter.

An optional #:kind specification chooses the decorative label, which defaults to "procedure". A #f result for kind-string-expr uses the default, otherwise kind-string-expr should produce a string. An alternate label should be all lowercase.

If #:id [src-id dest-id-expr] is supplied, then src-id is the identifier as it appears in the prototype (to be replaced by a defining instance), and dest-id-expr produces the identifier to be documented in place of src-id. This split between src-id and dest-id-expr roles is useful for functional abstraction of defproc.

Examples:
@defproc[(make-sandwich [ingredients (listof ingredient?)])
         sandwich?]{
  Returns a sandwich given the right ingredients.
}
 
@defproc[#:kind "sandwich-maker"
         (make-reuben [ingredient sauerkraut?] ...
                      [#:veggie? veggie? any/c #f])
         sandwich?]{
  Produces a reuben given some number of @racket[ingredient]s.
 
  If @racket[veggie?] is @racket[#f], produces a standard
  reuben with corned beef. Otherwise, produces a vegetable
  reuben.
}

syntax

(defproc* maybe-kind maybe-id
          ([prototype
            result-contract-expr-datum] ...)
          pre-flow ...)
Like defproc, but for multiple cases with the same id.

When an id has multiple calling cases, they must be defined with a single defproc*, so that a single definition point exists for the id. However, multiple distinct ids can also be defined by a single defproc*, for the case that it’s best to document a related group of procedures at once.

Examples:
@defproc[((make-pb&j)
          (make-pb&j [jelly jelly?]))
         sandwich?]{
  Returns a peanut butter and jelly sandwich. If @racket[jelly]
  is provided, then it is used instead of the standard (grape)
  jelly.
}

syntax

(defform maybe-kind maybe-id maybe-literals form-datum
  maybe-contracts
  pre-flow ...)
 
maybe-kind = 
  | #:kind kind-string-expr
     
maybe-id = 
  | #:id id
  | #:id [id id-expr]
     
maybe-literals = 
  | #:literals (literal-id ...)
     
maybe-contracts = 
  | 
#:contracts ([subform-datum contract-expr-datum]
             ...)
Produces a sequence of flow elements (encapsulated in a splice) to document a syntatic form named by id (or the result of id-expr) whose syntax is described by form-datum. If no #:id is used to specify id, then form-datum must have the form (id . datum).

If #:kind kind-string-expr is supplied, it is used in the same way as for defproc, but the default kind is "syntax".

If #:id [id id-expr] is supplied, then id is the identifier as it appears in the form-datum (to be replaced by a defining instance), and id-expr produces the identifier to be documented. This split between id and id-expr roles is useful for functional abstraction of defform.

The id (or result of id-expr) is indexed, and it is also registered so that racket-typeset uses of the identifier (with the same for-label binding) are hyperlinked to this documentation. The defmodule or declare-exporting requirements, as well as the binding requirements for id (or result of id-expr), are the same as for defproc.

The decoded pre-flow documents the form. In this description, a reference to any identifier in form-datum via racket, racketblock, etc. is typeset as a sub-form non-terminal. If #:literals clause is provided, however, instances of the literal-ids are typeset normally (i.e., as determined by the enclosing context).

If a #:contracts clause is provided, each subform-datum (typically an identifier that serves as a meta-variable in form-datum) is shown as producing a value that must satisfy the contract described by contract-expr-datum. Use #:contracts only to specify constraints on a value produced by an expression; for constraints on the syntax of a subform-datum, use grammar notation instead, possibly through an auxiliary grammar specified using defform/subs.

The typesetting of form-datum, subform-datum, and contract-expr-datum preserves the source layout, like racketblock.

Examples:
@defform[(sandwich-promise sandwich-expr)
         #:contracts ([sandwich-expr sandwich?])]{
  Returns a promise to construct a sandwich. When forced, the promise
  will produce the result of @racket[sandwich-expr].
}
 
@defform[#:literals (sandwich mixins)
         (sandwich-promise* [sandwich sandwich-expr]
                            [mixins ingredient-expr ...])
         #:contracts ([sandwich-expr sandwich?]
                      [ingreient-expr ingredient?])]{
  Returns a promise to construct a sandwich. When forced, the promise
  will produce the result of @racket[sandwich-expr]. Each result of
  the @racket[ingredient-expr]s will be mixed into the resulting
  sandwich.
}

syntax

(defform* maybe-kind maybe-id maybe-literals [form-datum ...+]
  maybe-contracts
  pre-flow ...)
Like defform, but for multiple forms using the same id.

Examples:
@defform*[((call-with-current-sandwich expr)
           (call-with-current-sandwich expr sandwich-handler-expr))]{
  Runs @racket[expr] and passes it the value of the current
  sandwich. If @racket[sandwich-handler-expr] is provided, its result
  is invoked when the current sandwich is eaten.
}

syntax

(defform/subs maybe-kind maybe-id maybe-literals form-datum
  ([nonterm-id clause-datum ...+] ...)
  maybe-contracts
  pre-flow ...)
Like defform, but including an auxiliary grammar of non-terminals shown with the id form. Each nonterm-id is specified as being any of the corresponding clause-datums, where the formatting of each clause-datum is preserved.

Examples:
@defform/subs[(sandwich-factory maybe-name factory-component ...)
              [(maybe-name (code:line)
                           name)
               (factory-component (code:line #:protein protein-expr)
                                  [vegetable vegetable-expr])]]{
  Constructs a sandwich factory. If @racket[maybe-name] is provided,
  the factory will be named. Each of the @racket[factory-component]
  clauses adds an additional ingredient to the sandwich pipeline.
}

syntax

(defform*/subs maybe-kind maybe-id maybe-literals [form-datum ...+]
  ([nonterm-id clause-datum ...+] ...)
  maybe-contracts
  pre-flow ...)
Like defform/subs, but for multiple forms for id.

syntax

(defform/none maybe-kind maybe-literal form-datum maybe-contracts
  pre-flow ...)
Like defform, but without registering a definition.

syntax

(defidform maybe-kind id pre-flow ...)

Like defform, but with a plain id as the form.

syntax

(defidform/inline id)

(defidform/inline (unsyntax id-expr))
Like defidform, but id (or the result of id-expr, analogous to defform) is typeset as an inline element. Use this form sparingly, because the typeset form does not stand out to the reader as a specification of id.

syntax

(specform maybe-literals datum maybe-contracts
  pre-flow ...)
Like defform, but without indexing or registering a definition, and with indenting on the left for both the specification and the pre-flows.

syntax

(specsubform maybe-literals datum maybe-contracts
  pre-flow ...)
Similar to defform, but without any specific identifier being defined, and the table and flow are typeset indented. This form is intended for use when refining the syntax of a non-terminal used in a defform or other specsubform. For example, it is used in the documentation for defproc in the itemization of possible shapes for arg-spec.

The pre-flows list is parsed as a flow that documents the procedure. In this description, a reference to any identifier in datum is typeset as a sub-form non-terminal.

syntax

(specsubform/subs maybe-literals datum
  ([nonterm-id clause-datum ...+] ...)
  maybe-contracts
  pre-flow ...)
Like specsubform, but with a grammar like defform/subs.

syntax

(specspecsubform maybe-literals datum maybe-contracts
  pre-flow ...)
Like specsubform, but indented an extra level. Since using specsubform within the body of specsubform already nests indentation, specspecsubform is for extra indentation without nesting a description.

syntax

(specspecsubform/subs maybe-literals datum
 ([nonterm-id clause-datum ...+] ...)
 maybe-contracts
 pre-flow ...)
Like specspecsubform, but with a grammar like defform/subs.

syntax

(defparam id arg-id contract-expr-datum pre-flow ...)

Like defproc, but for a parameter. The contract-expr-datum serves as both the result contract on the parameter and the contract on values supplied for the parameter. The arg-id refers to the parameter argument in the latter case.

Examples:
@defparam[current-sandwich sandwich sandwich?]{
  A parameter that defines the current sandwich for operations that
  involve eating a sandwich.
}

syntax

(defboolparam id arg-id pre-flow ...)

Like defparam, but the contract on a parameter argument is any/c, and the contract on the parameter result is boolean?.

syntax

(defthing maybe-kind id contract-expr-datum pre-flow ...)

Like defproc, but for a non-procedure binding.

If #:kind kind-string-expr is supplied as maybe-kind, it is used in the same way as for defproc, but the default kind is "value".

Examples:
@defthing[moldy-sandwich sandwich?]
  Don't eat this. Provided for backwards compatibility.
}

syntax

(defstruct* struct-name ([field-name contract-expr-datum] ...)
  maybe-mutable maybe-non-opaque maybe-constructor
  pre-flow ...)

syntax

(defstruct struct-name ([field-name contract-expr-datum] ...)
  maybe-mutable maybe-non-opaque maybe-constructor
  pre-flow ...)
 
struct-name = id
  | (id super-id)
     
maybe-mutable = 
  | #:mutable
     
maybe-non-opaque = 
  | #:prefab
  | #:transparent
     
maybe-constructor = 
  | #:constructor-name constructor-id
  | #:extra-constructor-name constructor-id
Similar to defform or defproc, but for a structure definition. The defstruct* form corresponds to struct, while defstruct corresponds to define-struct.

Examples:
@defstruct[sandwich ([protein ingredient?] [sauce ingredient?])]{
  A strucure type for sandwiches. Sandwiches are a pan-human foodstuff
  composed of a partially-enclosing bread material and various
  ingredients.
}

syntax

(deftogether [def-expr ...] pre-flow ...)

Combines the definitions created by the def-exprs into a single definition box. Each def-expr should produce a definition point via defproc, defform, etc. Each def-expr should have an empty pre-flow; the decoded pre-flow sequence for the deftogether form documents the collected bindings.

Examples:
@deftogether[(@defthing[test-sandwich-1 sandwich?]
              @defthing[test-sandwich-2 sandwich?])]{
  Two high-quality sandwiches. These are provided for convenience
  in writing test cases
}

syntax

(racketgrammar maybe-literals id clause-datum ...+)

 
maybe-literals = 
  | #:literals (literal-id ...)
Creates a table to define the grammar of id. Each identifier mentioned in a clause-datum is typeset as a non-terminal, except for the identifiers listed as literal-ids, which are typeset as with racket.

syntax

(racketgrammar* maybe-literals [id clause-datum ...+] ...)

Like racketgrammar, but for typesetting multiple productions at once, aligned around the = and |.

procedure

(defidentifier id    
  [#:form? form?    
  #:index? index?    
  #:show-libs? show-libs?])  element?
  id : identifier?
  form? : boolean? = #f
  index? : boolean? = #t
  show-libs? : boolean? = #t
Typesets id as a Racket identifier, and also establishes the identifier as the definition of a binding in the same way as defproc, defform, etc. As always, the library that provides the identifier must be declared via defmodule or declare-exporting for an enclosing section.

If form? is a true value, then the identifier is documented as a syntactic form, so that uses of the identifier (normally including id itself) are typeset as a syntactic form.

If index? is a true value, then the identifier is registered in the index.

If show-libs? is a true value, then the identifier’s defining module may be exposed in the typeset form (e.g., when viewing HTML and the mouse hovers over the identifier).

syntax

(schemegrammar maybe-literals id clause-datum ...+)

syntax

(schemegrammar* maybe-literals [id clause-datum ...+] ...)

Compatibility aliases for racketgrammar and racketgrammar*.

Specifies the target maximum width in characters for the output of defproc and defstruct.