Data Types

Basic Data Types

L4's basic data types are

• Booleans

• strings

• numbers

• records

Booleans

L4's Boolean values are TRUE and FALSE. The type of Booleans is called BOOLEAN.

Strings

Strings are quoted using double-quotes: "Alice Avocado" is a string. (Note that `Alice Avocado` is not a string, it is an identifer with a compound name, and if used, its meaning must be defined somewhere in the program.) The type of strings is called STRING.

Numbers

L4 does not currently distinguish between integers and floats and just has a common type called NUMBER.

Literal numbers can be written as usual: 333, 0.25 and -77.2 are examples of valid numbers.

User-Defined Types

Enumerations, or Enums

An enumeration type can be declared as follows:

DECLARE Colour IS ONE OF red
                         green
						 blue

This introduces a new type Colour (distinct from all other types) with exactly three values, which are called red, green and blue.

If you try to use purple as a Colour, it will be rejected.

Destructuring with CONSIDER

We pattern-match the possible values of a variable using CONSIDER / WHEN / THEN:

GIVEN c IS A Colour
DECIDE `is qing` IF
   CONSIDER c
       WHEN red   THEN FALSE
       WHEN green THEN TRUE
       WHEN blue  THEN TRUE

In classical Chinese, blue is 青, but green is also 青.

https://www.lomography.com/magazine/337259-color-chronicles-deconstructing-qing

Type Aliases

We can declare our own types in terms of the above basic types:

DECLARE `count of fruit in my bag` IS A NUMBER
DECLARE `common name of fruit`     IS A STRING
DECLARE `scientific name of fruit` IS A STRING
DECLARE `my bag has a hole`        IS A BOOLEAN

This creates aliases from a user-defined type to a native type. Aliases are exchangeable with the type they expand to, their primary intent is to provide more descriptive names in specific situations.

Records

Records in L4 are used to group related data together.

Record Definition

Record types are declared using DECLARE ... HAS ... IS A ... syntax.

DECLARE Person
  HAS name IS A STRING
      age  IS A NUMBER

This introduces a new type Person (distinct from all other types) comprising two fields, one called name being a string, and one called age being a number.

Once the type has been declared, you can instantiate it by defining variables using MEANS syntax.

There are two ways to give the attributes.

Using WITH ... IS ... syntax:

alice MEANS
  Person WITH name IS "Alice Avocado"
              age  IS 21

If using WITH ... IS ... syntax, argument order is flexible, so you could also define:

bob MEANS
  Person WITH age  IS 20
              name IS "Bob Blueberry"

One can also use the shorter Type OF ... syntax:

alice MEANS Person OF "Alice Avocado", 21

In this case, the arguments have to appear exactly in the order they have been listed in the declaration of the type Person above, so the name has to come before the age.

You can omit the OF when the attributes go down the page:

alice MEANS Person   "Alice Avocado"
                     21

Accessing Record Fields

Fields of a record can be accessed using the possessive apostrophe-s ('s) notation:

GIVEN john IS A Person
GIVETH A STRING
johnsName MEANS john's name

GIVETH A NUMBER
johnsAge MEANS john's age

This is analogous to the use of a dot (.) used in most other languages for record accessors: john.age becomes john's age

Algebraic data types

Some types are neither enumeration nor record types, but a combination of the two. (TODO.)

Advanced Concepts

More on these later.

Type-Directed Name Resolution (TDNR)

TDNR allows the same identifier to be in scope multiple times with different types.

ASSUME foo IS A NUMBER
ASSUME foo IS A BOOLEAN
ASSUME foo IS A STRING