The examples make use of the following type declarations:
:- type type1
---> type1(
field1 :: type2,
field2 :: string
).
:- type type2
---> type2(
field3 :: int,
field4 :: int
).
The compiler generates some field access functions for ‘field1’. The functions generated for the other fields are similar.
:- func type1 ^ field1 = type2. type1(Field1, _) ^ field1 = Field1. :- func (type1 ^ field1 := type2) = type1. (type1(_, Field2) ^ field1 := Field1) = type1(Field1, Field2).
Using these functions and the syntactic sugar described in Record syntax, programmers can write code such as
:- func type1 ^ increment_field3 = type1.
Term0 ^ increment_field3 =
Term0 ^ field1 ^ field3 := Term0 ^ field1 ^ field3 + 1.
The compiler expands this into
increment_field3(Term0) = Term :-
OldField3 = field3(field1(Term0)),
OldField1 = field1(Term0),
NewField1 = 'field3 :='(OldField1, OldField3 + 1),
Term = 'field1 :='(Term0, NewField1).
The field access functions defined in the Mercury standard library module ‘map’ can be used as follows:
:- func update_field_in_map(map(int, type1), int, string)
= map(int, type1) is semidet.
update_field_in_map(Map, Index, Value) =
Map ^ elem(Index) ^ field2 := Value.