CLI language of created by Microsoft Research
Maintained and improved by the F# Software Foundation
"F# is a mature, open source, cross-platform, functional-first programming language.
It empowers users and organizations to tackle complex computing problems with simple, maintainable and robust code"
mature : released in 2005, latest version is F# 4.1. MSDN docs available F#. Microsoft supports it as C#. Available since VS 2012 (at least)
functional-first : not limited to functional paradigm. Possible to mix to object oriented or procedural code
(* binding *)
let amount = 10
(* declare function and conditional *)
let incrementWithLimit number limit =
if number < limit then
number + 1
else
limit
(* reference & array & loop *)
open System.IO
let folders = Directory.GetDirectories @"C:\"
for folder in folders do
printfn "Folder found in C:\\ : %s" folder
(* n-tuples *)
let filmAndScore = ("inception", 8.8)
let filmScoreYear = ("interstellar", 8.6, 2014)
Compiler deduces types.
This information is not hidden. IDEs integrating F# provide it
let isAwesome = true (* bool *)
let increment n = n + 1 (* (int -> int) *)
let increment n = n + 1. (* (float -> float) *)
Compiler notifies of ambiguous cases. We fix this by clarifying the expected type : type annotations
(* ('a -> 'b) *)
let getLength x = x.Length
(* (string -> int) *)
let getLength (x:string) = x.Length
(* (float[] -> int) *)
let getLength (x:float []) = x.Length
Value is bound to the label, forever
Limits surprise effects by working with copies, thread safety
BTW System.String is immutable... and you're still alive
Do not see it as a negative point
functions accepts or return other functions as parameters
let sendEmail recipient recipientCheck =
recipientCheck recipient
The power to build "pre configured" functions
(* t_credentials -> t_query -> t_result *)
let connectToDbThenExecute credentials query =
DB.connect credentials
DB.exec query
connectToDbThenExecute creds query1
connectToDbThenExecute creds query2
let executeQuery = connectToDbThenExecute creds
executeQuery query1
executeQuery query2
How do you represent a set of exclusive cases ?
/// booleans
var isUserAvailable = true
/// 0 stands for rock, 1 for paper,
/// 2 for scissors.
var player1Choice = 1
///... until the days someone adds a dynamite case
/// and "explodes" in your code
/// Dynamite case introduced.
/// Effective against scissors and paper
var player2Choice = 4
var isScissors = true;
var isRock = false;
var isPaper = false;
var isDynamite = true;
public enum TurnChoice
{
Player1,
Player2
}
/// ... has to evolve into ...
public enum AdvancedTurnChoice
{
Player1Day,
Player1Night,
Player2Day,
Player2Night,
}
Sometimes, we have constraints or we change the design as if nothing happened and we hold limited data
~ Union in C
type UserChoice =
| Paper
| Scissors
| Rock
let user1Choice = Paper
(* so enums ? no because I did not write: *)
type UserChoice =
| Paper = 0
| Scissors = 1
| Rock = 2
type FamousCaseOfShapeModelization =
| Circle of uint32 (* ( ͡° ͜ʖ ͡°) *)
| Square of Edge : uint32
| Rectangle of uint32 * uint32
let TimesSquare = Square(12)
Multiple DU usage
type PlayingCardKind =
| Clubs | Spades | Diamonds | Hearts
type PlayingCardNum =
| Ace | King | Queen | Jack | Two | Three | Four
| Five | Six | Seven | Eight | Nine | Ten
type PlayingCard =
| PlayingCardNum * PlayingCardKind
type Deck = PlayingCard []
We could use enums ... Now imagine card values as floats...
wrong/illegal states are not possible by design
type BinaryTree =
| Node of 'NodeType * Tree * Tree
| Leaf of 'NodeType
if then else : conditional
This construct lets you branch between multiple expressions to execute with the ability to decompose or match input data
match data with
| case 1 -> expression 1
| ...
| case N -> expression N
One benefit is that missing cases are detected by the compiler.
DUs are supported
so you will know at compile time that your design is broken
match cardKind with
| Clubs -> "♣"
| Spades -> "♠"
| Diamonds -> "♦" (* no hearts, won't compile *)
Data extraction is possible by creating a "pattern"
let todayActivities = ["curling"; "swimming"; "rock climbing"]
match todayActivities with
| "curling"::remainingElements -> "Excellent activity !"
| ...
type Meal = Breakfast | Lunch | Dinner
let lunchAndPriceTuple = (Breakfast, 15) (* Meal*int *)
let mealGossip tpl =
match mealTuple with
| (Breakfast, price) when price > 8 -> "omg so pricey!"
| (Breakfast, price) -> "where did you buy it ?"
| (Lunch, price) ...
Another level than if/then/else (we could even replace them)
Improved readability
What is the workflow when working with data of an external source ?
=>Read the doc, analyse response, create types, play with properties
How to work with data from external source in short cycles (schema changes ?) and with a need to be operational quickly ?
=>By using the same workflow ?
The concept is to generate on the fly the structures and properties to explore and use them.
We may miss information in the doc. Which tool could let us discover it ?
[<LiteralAttribute>] (* compile time assignment *)
let url = "https://jsonplaceholder.typicode.com/users"
type LectureDataProvider = JsonProvider<url>
let allUsers = LectureDataProvider.Load(url)
Wrap up
The difficulty is to set the cursor between FP and F# so some stuff is left
Not covered in this presentation because of time: