elm-bounded-nat

Type-safe natural numbers → >= 0 can ensure that a Nat is in a given range at compile-time.

toHexChar : Nat (ArgIn min_ Nat15 ifN_) -> Char

No number below 0 or above 15 can be passed in as an argument!

examples

elm install lue-bird/elm-typed-value
elm install lue-bird/elm-bounded-nat

import Nat exposing (Nat, Min, In, N, Is, To, ArgIn)
import NNats exposing (..)
    -- (..) is nat0 to nat160
import InNat
import MinNat

import TypeNats exposing (..)
    -- (..) is Nat0 to Nat160 & Nat1Plus to Nat160Plus

import Typed exposing (val, val2)

color

rgb : Float -> Float -> Float -> Color

This is common, but

• the one implementing the function has to handle the cases where a value is not between 0 and 1
• the type doesn't tell us that a Float between 0 & 1 is wanted

rgbPer100 :
    Nat (ArgIn redMin_ Nat100 redIfN_)
    -> Nat (ArgIn greenMin_ Nat100 greenIfN_)
    -> Nat (ArgIn blueMin_ Nat100 blueIfN_)
    -> Color

• the one using the function must prove that the numbers are actually between 0 and 100
• you clearly know what input is desired

The type

Nat (ArgIn min_ Nat100 ifN_)

is saying it wants:

an integer >= 0                  Nat          
  in a range                        ArgIn       
    at least any minimum value        min_
    at most 100                       Nat100
    which might be exact              ifN_

They can prove it by

• using exact values

red = rgbPer100 nat100 nat0 nat0 -- 👍

nat0 : Nat (N Nat0 atLeast0_ ...)
-- between 0 and 0/1/.../100(/101/...)

nat100 : Nat (N Nat100 (N100Plus orMore_) ...)
-- between 100 and 100(/101/...)

• handling the possibility that a number isn't in the expected range

toPositive : Int -> Maybe (Nat (Min Nat1))
toPositive =
    Nat.isIntAtLeast nat1

• clamping

grey float =
    let
        greyLevel =
            float
                * 100
                |> round
                |> Nat.intInRange nat0 nat100
    in
    rgbPer100 greyLevel greyLevel greyLevel

• There are more ways, but you get the idea 🙂

digit

toDigit : Char -> Maybe Int

You might be able to do anything with this Int value, but you lost useful information.

• Can the result even be negative?
• Can the result even have multiple digits?

toDigit : Char -> Maybe Digit

type alias Digit =
    Nat (In Nat0 Nat9)

The type of a value reflects how much you know.

• In: between a minimum & maximum value
• Min: at least a minimum value
• N: exact value
  • also describes the difference between 2 values

factorial

intFactorial : Int -> Int
intFactorial x =
    if x == 0 then
        1

    else
        x * intFactorial (x - 1)

This forms an infinite loop if we call intFactorial -1...

Let's disallow negative numbers here!

factorial : Nat (ArgIn min_ max_ ifN_) -> Nat (Min Nat1)
factorial =
    factorialBody

Says: for every natural number n >= 0, n! >= 1.

factorialBody : -- as in factorial
factorialBody =
    case x |> MinNat.isAtLeast nat1 { lowest = nat0 } of
        Nat.Below _ ->
            MinNat.value nat1

        Nat.EqualOrGreater atLeast1 ->
            -- atLeast1 is a Nat (Min Nat1)
            Nat.mul atLeast1
                (factorial
                    (atLeast1 |> MinNat.sub nat1)
                    -- so we can subtract 1
                )

factorial nat4 --> Nat 24

→ There is no way to put a negative number in.

→ We have the extra promise, that every result is >= 1

Sadly, we need seperate factorial & factorialBody because of a compiler bug.

But we can do even better! !19 is already > the maximum safe Int 2^53 - 1.

safeFactorial : Nat (ArgIn min_ Nat18 ifN_) -> Nat (Min Nat1)
safeFactorial =
    factorial

No extra work.

tips

• keep as much type information as possible and drop it only where you need to.

squares2To10 =
    Nat.range nat2 nat10
        -- more info than List.range 2 10
        -- → every Nat is In Nat2 (Nat10Plus a_)
        |> List.map
            (Nat.toPower nat2
                -- we still know it's >= 2
            )

• keep your function annotations as general as possible

Instead of accepting only exact values

rgb : Nat (N red_ atLeastRed_ (Is redTo100_ To Nat100) redIs_) -> --...

accept values that are somewhere in a range.

rgb : Nat (ArgIn redMin_ Nat100 ifN_) -> --...

ifN_ says that it can be exact anyway. Or instead of

charFromCode : Nat (Min min_) -> Char

which you should also never do, allow Nat (In min ...) with any max & Nat (N ...) to fit in as well!

charFromCode : Nat (ArgIn min_ max_ ifN_) -> Char

Take a look at typesafe-array to see a lot of this in action!

You get to know that

• a Nat (ArgIn ...) is very useful as an index
• In, Min, Only can also describe the array length