colon-equals: Define a new data type from a sequence of constructors.

In mailund/pmatch: Pattern Matching

Description

This assignment operator introduces a domain-specific language for specifying new types. Types are defined by the ways they can be constructed. This is provided as a sequence of |-separated constructors, where a constructor is either a constant, i.e., a bare symbol, or a function.

Usage

`:=`(data_type, constructors)

Arguments

data_type	The name of the new data type. Should be given as a bare symbol.
constructors	A list of |-separated constructor specifications.

Details

We can construct an enumeration like this:

numbers := ONE | TWO | THREE

This will create the type numbers and three constants, ONE, TWO, and THREE that can be matched against using the case_func function.

f <- case_func(ONE -> 1, TWO -> 2, THREE -> 3) x <- TWO f(x)

Evaluating functions declared using the case_func function will compare the value in x against the three patterns and recognize that x holds the constant TWO and it will then return 2.

With function constructors we can create more interesting data types. For example, we can create a linked list like this

linked_list := NIL | CONS(car, cdr : linked_list)

This expression defines constant NIL and function CONS. The function takes two arguments, car and cdr, and requires that cdr has type linked_list. We can create a list with three elements, 1, 2, and 3, by writing

CONS(1, CONS(2, CONS(3, NIL)))

and we can, e.g., test if a list is empty using

f <- case_func(NIL -> TRUE, CONS(car,cdr) -> FALSE) f(lst)

A variable will be bound to any value, and you can exploit this to get a default return value. I prefer to use .

f <- case_func(NIL -> TRUE, . -> FALSE) f(lst)

Arguments to a constructor function can be typed. To specify typed variables, we use the :-operator. The syntax is then var : type. The type will be checked when you construct a value using the constructor.

Examples

linked_list := NIL | CONS(car, cdr:linked_list)
lst <- CONS(1, CONS(2, CONS(3, NIL)))
len <- case_func(
    acc = 0,
    NIL -> acc,
    CONS(car, cdr) -> len(cdr, acc + 1)
)
len(lst)

list_sum <- case_func(
    acc = 0,
    NIL -> acc,
    CONS(car, cdr) -> list_sum(cdr, acc + car)
)
list_sum(lst)

mailund/pmatch documentation built on Feb. 4, 2020, 4:55 a.m.