Skip to content

C# functional language extensions and 'Erlang like' concurrency system

License

Notifications You must be signed in to change notification settings

pdegenhardt/language-ext

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

lang-ext

C# Functional Language Extensions

Join the chat at https://gitter.im/louthy/language-ext

This library uses and abuses the features of C# 6 to provide a functional 'Base class library', that, if you squint, can look like extensions to the language itself. It also includes an 'Erlang like' process system (actors) that can optionally persist messages and state to Redis (note you can use it without Redis for in-app messaging). The process system additionally supports Rx streams of messages and state allowing for a complete system of reactive events and message dispatch.

Nu-get package Description
LanguageExt.Core All of the core types and functional 'prelude'. This is all that's needed to get started.
LanguageExt.FSharp F# to C# interop library. Provides interop between the LanguageExt.Core types (like Option, List and Map) to the F# equivalents, as well as interop between core BCL types and F#
LanguageExt.Parsec Port of the Haskell parsec library
LanguageExt.Process 'Erlang like' actor system for in-app messaging and massive concurrency
LanguageExt.Process.Redis Cluster support for the LanguageExt.Process system for cluster aware processes using Redis for queue and state persistence
LanguageExt.Process.FSharp F# API to the LanguageExt.Process system
LanguageExt.ProcessJS Javascript API to the LanguageExt.Process system. Supports running of Processes in a client browser, with hooks for two-way UI binding.

Twitter: https://twitter.com/paullouth

Introduction

One of the great new features of C# 6 is that it allows us to treat static classes like namespaces. This means that we can use static methods without qualifying them first. This instantly gives us access to single term method names that look exactly like functions in functional languages. i.e.

    using static System.Console;
    
    WriteLine("Hello, World");

This library tries to bring some of the functional world into C#. It won't always sit well with the seasoned C# OO programmer, especially the choice of camelCase names for a lot of functions and the seeming 'globalness' of a lot of the library.

I can understand that much of this library is non-idiomatic; But when you think of the journey C# has been on, is idiomatic necessarily right? A lot of C#'s idioms are inherited from Java and C# 1.0. Since then we've had generics, closures, Func, LINQ, async... C# as a language is becoming more and more like a functional language on every release. In fact the bulk of the new features are either inspired by or directly taken from features in functional languages. So perhaps it's time to move the C# idioms closer to the functional world's idioms?

A note about naming

One of the areas that's likely to get seasoned C# heads worked up is my choice of naming style. The intent is to try and make something that 'feels' like a functional language rather than follow the 'rule book' on naming conventions (mostly set out by the BCL).

There is however a naming guide that will stand you in good stead whilst reading through this documentation:

  • Type names are PascalCase in the normal way
  • The types all have constructor functions rather than public constructors that you new. They will always be PascalCase:
    Option<int> x = Some(123);
    Option<int> y = None;
    List<int> items = List(1,2,3,4,5);
    Map<int,string> dict = Map(Tuple(1, "Hello"), Tuple(2, "World"));
  • Any (non-type constructor) static functions that can be used on their own by using static LanguageExt.Prelude are camelCase.
    var x = map(opt, v => v * 2);
  • Any extension methods, or anything 'fluent' are PascalCase in the normal way
    var x = opt.Map(v => v * 2);

Even if you don't agree with this non-idiomatic approach, all of the camelCase static functions have fluent variants, so actually you never have to see the 'non-standard' stuff.

If you're not using C# 6 yet, then you can still use this library. Anywhere in the docs below where you see a camelCase function it can be accessed by prefixing with Prelude.

Getting started

To use this library, simply include LanguageExt.Core.dll in your project or grab it from NuGet. And then stick this at the top of each cs file that needs it:

using LanguageExt;
using static LanguageExt.Prelude;

The namespace LanguageExt contains the types, and LanguageExt.Prelude contains the functions.

There is also:

(more on those later)

Features

This library is quickly becoming a 'Base Class Library' for functional programming in C#. The features include:

Location Feature Description
Core Lst<T> Immutable list
Core Map<K,V> Immutable map
Core Set<T> Immutable set
Core Que<T> Immutable queue
Core Stck<T> Immutable stack
Core Option<T> Option monad that can't be used with null values
Core OptionUnsafe<T> Option monad that can be used with null values
Core Either<L,R> Right/Left choice monad that won't accept null values
Core EitherUnsafe<L,R> Right/Left choice monad that can be used with null values
Core Try<T> Exception handling lazy monad
Core TryOption<T> Option monad with third state 'Fail' that catches exceptions
Core Reader<E,T> Reader monad
Core Writer<O,T> Writer monad
Core State<S,T> State monad
Core Rws<E,O,S,T> Reader/Writer/State monad
Parsec Parser<T> Parser monad and full parser combinators library
Core NewType<T> Haskell newtype equivalent i.e: class Hours : NewType<double> { public Hours(double value) : base(value) { } }. The resulting type is: equatable, comparable, foldable, a functor, monadic, appendable, subtractable, divisible, multiplicable, and iterable
Core Nullable<T> extensions Extension methods for Nullable<T> that make it into a functor, applicative, foldable, iterable and a monad
Core Task<T> extensions Extension methods for Task<T> that make it into a functor, applicative, foldable, iterable and a monad
Core Monad transformers A higher kinded type (ish)
Process Process library Actor system. The same as Erlang processes for massive concurrency with state management.
Process.Redis Redis persistence Persistence of the Process system message-queues and state, for robustness and inter-app communication.
Core Currying [Translate the evaluation of a function that takes multiple arguments into a sequence of functions, each with a single argument](https://louthy.github.io/languageext.core/LanguageExt/Prelude_.htm#curry<T1, T2, R>)
Core Partial application [the process of fixing a number of arguments to a function, producing another function of smaller arity](https://louthy.github.io/languageext.core/LanguageExt/Prelude_.htm#par<T1, T2, R>)
Core Memoization [An optimization technique used primarily to speed up programs by storing the results of expensive function calls and returning the cached result when the same inputs occur again](https://louthy.github.io/languageext.core/LanguageExt/Prelude_.htm#memo<T, R>)
Core Improved lambda type inference var add = fun( (int x, int y) => x + y)
Core IObservable<T> extensions

It started out trying to deal with issues in C#, that after using Haskell and F# started to frustrate me:

  • Poor tuple support
  • Null reference problem
  • Lack of lambda and expression inference
  • Void isn't a real type
  • Mutable lists and dictionaries
  • The awful 'out' parameter

Poor tuple support

I've been crying out for proper tuple support for ages. It looks like we're no closer with C# 6. The standard way of creating them is ugly Tuple.Create(foo,bar) compared to functional languages where the syntax is often (foo,bar) and to consume them you must work with the standard properties of Item1...ItemN. No more...

    var ab = Tuple("a","b");

Now isn't that nice?

Consuming the tuple is now handled using Map, which projects the Item1...ItemN onto a lambda function (or action):

    var name = Tuple("Paul","Louth");
    var res = name.Map( (first,last) => String.Format("{0} {1}", first, last) );

Or, you can use a more functional approach:

    var name = Tuple("Paul","Louth");
    var res = map( name, (first,last) => String.Format("{0} {1}", first, last) );

This allows the tuple properties to have names, and it also allows for fluent handling of functions that return tuples.

Null reference problem

null must be the biggest mistake in the whole of computer language history. I realise the original designers of C# had to make pragmatic decisions, it's a shame this one slipped through though. So, what to do about the 'null problem'?

null is often used to indicate 'no value'. i.e. the method called can't produce a value of the type it said it was going to produce, and therefore it gives you 'no value'. The thing is that when 'no value' is passed to the consuming code, it gets assigned to a variable of type T, the same type that the function said it was going to return, except this variable now has a timebomb in it. You must continually check if the value is null, if it's passed around it must be checked too.

As we all know it's only a matter of time before a null reference bug crops up because the variable wasn't checked. It puts C# in the realm of the dynamic languages, where you can't trust the value you're being given.

Functional languages use what's known as an 'option type'. In F# it's called Option in Haskell it's called Maybe. In the next section we'll see how it's used.

Option

Option<T> works in a very similar way to Nullable<T> except it works with all types rather than just value types. It's a struct and therefore can't be null. An instance can be created by either calling Some(value), which represents a positive 'I have a value' response; Or None, which is the equivalent of returning null.

So why is it any better than returning T and using null? It seems we can have a non-value response again right? Yes, that's true, however you're forced to acknowledge that fact, and write code to handle both possible outcomes because you can't get to the underlying value without acknowledging the possibility of the two states that the value could be in. This bulletproofs your code. You're also explicitly telling any other programmers that: "This method might not return a value, make sure you deal with that". This explicit declaration is very powerful.

This is how you create an Option<int>:

    var optional = Some(123);

To access the value you must check that it's valid first:

    int x = optional.Match( 
                Some: v  => v * 2,
                None: () => 0 
                );

An alternative (functional) way of matching is this:

    int x = match( optional, 
                   Some: v  => v * 2,
                   None: () => 0 );

Yet another alternative (fluent) matching method is this:

    int x = optional
               .Some( v  => v * 2 )
               .None( () => 0 );

So choose your preferred method and stick with it. It's probably best not to mix styles.

There are also some helper functions to work with default None values, You won't see a .Value or a GetValueOrDefault() anywhere in this library. It is because .Value puts us right back to where we started, you may as well not use Option<T> in that case. GetValueOrDefault() is as bad, because it can return null for reference types, and depending on how well defined the struct type is you're working with: a poorly defined value type.

However, clearly there will be times when you don't need to do anything with the Some case, because, well that's what you asked for. Also, sometimes you just want some code to execute in the Some case and not the None case...

    // Returns the Some case 'as is' and 10 in the None case
    int x = optional.IfNone(10);        

    // As above, but invokes a Func<T> to return a valid value for x
    int x = optional.IfNone(() => GetAlternative());        
    
    // Invokes an Action<T> if in the Some state.
    optional.IfSome(x => Console.WriteLine(x));

Of course there are functional versions of the fluent version above:

    int x = ifNone(optional, 10);
    int x = ifNone(optional, () => GetAlternative());
    ifSome(optional, x => Console.WriteLine(x));

To smooth out the process of returning Option<T> types from methods there are some implicit conversion operators and constructors:

    // Implicitly converts the integer to a Some of int
    Option<int> GetValue()
    {
        return 1000;
    }

    // Implicitly converts to a None of int
    Option<int> GetValue() => 
    {
        return None;
    }
    
    // Will handle either a None or a Some returned
    Option<int> GetValue(bool select) =>
        select
            ? Some(1000)
            : None;
            
    // Explicitly converts a null value to None and a non-null value to Some(value)
    Option<string> GetValue()
    {
        string value = GetValueFromNonTrustedApi();
        return Optional(value);
    }
            
    // Implicitly converts a null value to None and a non-null value to Some(value)
    Option<string> GetValue()
    {
        string value = GetValueFromNonTrustedApi();
        return value;
    }

It's actually nearly impossible to get a null out of a function, even if the T in Option<T> is a reference type and you write Some(null). Firstly it won't compile, but you might think you can do this:

    private Option<string> GetStringNone()
    {
        string nullStr = null;
        return Some(nullStr);
    }

That will compile, but at runtime will throw a ValueIsNullException. If you do either of these (below) you'll get a None.

    private Option<string> GetStringNone()
    {
        string nullStr = null;
        return nullStr;
    }

    private Option<string> GetStringNone()
    {
        string nullStr = null;
        return Optional(nullStr);
    }

These are the coercion rules:

Converts from Converts to
x Some(x)
null None
None None
Some(x) Some(x)
Some(null) ValueIsNullException
Some(None) Some(None)
Some(Some(x)) Some(Some(x))
Some(Nullable null) ValueIsNullException
Some(Nullable x) Some(x)
Optional(x) Some(x)
Optional(null) None
Optional(Nullable null) None
Optional(Nullable x) Some(x)

As well as the protection of the internal value of Option<T>, there's protection for the return value of the Some and None handler functions. You can't return null from those either, an exception will be thrown.

    // This will throw a ResultIsNullException exception
    string res = GetValue(true)
                     .Some(x => (string)null)
                     .None((string)null);

So null goes away if you use Option<T>.

However, there are times when you want your Some and None handlers to return null. This is mostly when you need to use something in the BCL or from a third-party library, so momentarily you need to step out of your warm and cosy protected optional bubble, but you've got an Option<T> that will throw an exception if you try.
So you can use matchUnsafe and ifNoneUnsafe:

    string x = matchUnsafe( optional,
                            Some: v => v,
                            None: () => null );

    string x = ifNoneUnsafe( optional, null );
    string x = ifNoneUnsafe( optional, () => GetNull() );

And fluent versions:

    string x = optional.MatchUnsafe(
                   Some: v => v,
                   None: () => null 
                   );
    string x = optional.IfNoneUnsafe(null);
    string x = optional.IfNoneUnsafe(() => GetNull());

That is consistent throughout the library. Anything that could return null has the Unsafe suffix. That means that in those unavoidable circumstances where you need a null, it gives you and any other programmers working with your code the clearest possible sign that they should treat the result with care.

Option monad - gasp! Not the M word!

I know, it's that damn monad word again. They're actually not scary at all, and damn useful. But if you couldn't care less (or could care less, for my American friends), it won't stop you taking advantage of the Option<T> type. However, Option<T> type also implements Select and SelectMany and is therefore monadic. That means it can be used in LINQ expressions, but it means much more also.

    Option<int> two = Some(2);
    Option<int> four = Some(4);
    Option<int> six = Some(6);
    Option<int> none = None;

    // This expression succeeds because all items to the right of 'in' are Some of int
    // and therefore it lands in the Some lambda.
    int r = match( from x in two
                   from y in four
                   from z in six
                   select x + y + z,
                   Some: v => v * 2,
                   None: () => 0 );     // r == 24

    // This expression bails out once it gets to the None, and therefore doesn't calculate x+y+z
    // and lands in the None lambda
    int r = match( from x in two
                   from y in four
                   from _ in none
                   from z in six
                   select x + y + z,
                   Some: v => v * 2,
                   None: () => 0 );     // r == 0

This can be great for avoiding the use of if then else, because the computation continues as long as the result is Some and bails otherwise. It is also great for building blocks of computation that you can compose and reuse. Yes, actually compose and reuse, not like OO where the promise of composability and modularity are essentially lies.

To take this much further, all of the monads in this library implement a standard 'functional set' of functions:

    Sum                 // For Option<int> it's the wrapped value.
    Count               // For Option<T> is always 1 for Some and 0 for None.  
    Bind                // Part of the definition of anything monadic - SelectMany in LINQ
    Exists              // Any in LINQ - true if any element fits a predicate
    Filter              // Where in LINQ
    Fold                // Aggregate in LINQ
    ForAll              // All in LINQ - true if all element(s) fits a predicate
    Iter                // Passes the wrapped value(s) to an Action delegate
    Map                 // Part of the definition of any 'functor'.  Select in LINQ
    Lift / LiftUnsafe   // Different meaning to Haskell, this returns the wrapped value.  Dangerous, should be used sparingly.
    Select
    SeletMany
    Where

This makes them into what would be known in Haskell as a Type Class (although more of a catch-all type-class than a set of well-defined type-classes).

Monad transformers

Now the problem with C# is it can't do higher order polymorphism (imagine saying Monad<Option<T>> instead of Option<T>, Either<L,R>, Try<T>, IEnumerable<T>. And then the resulting type having all the features of the Option as well as the standard interface to Monad).

There's a kind of cheat way to do it in C# through extension methods. It still doesn't get you a single type called Monad<T>, so it has limitations in terms of passing it around. However it makes some of the problems of dealing with nested monadic types easier.

For example, below is a list of optional integers: Lst<Option<int>> (see lists later). We want to double all of the Some values, leave the None alone and keep everything in the list:

    using LanguageExt.Trans;  // Required for all transformer extension methods

    var list = List(Some(1), None, Some(2), None, Some(3));
    
    var presum = list.SumT();                                // 6
    
    list  = list.MapT( x => x * 2 );
    
    var postsum = list.SumT();                               // 12

Notice the use of MapT instead of Map (and SumT instead of Sum). If we used Map (equivalent to Select in LINQ), it would look like this:

    var list  = List(Some(1), None, Some(2), None, Some(3));
    
    var presum = list.Map(x => x.Sum()).Sum();
    
    list = list.Map( x => x.Map( v => v * 2 ) );
    
    var postsum = list.Map(x => x.Sum()).Sum();

As you can see the intention is much clearer in the first example. And that's the point with functional programming most of the time. It's about declaring intent rather than the mechanics of delivery.

To make this work we need extension methods for List<Option<T>> that define MapT and SumT [for the one example above]. And we need one for every pair of monads in this library (for one level of nesting A<B<T>>), and for every function from the 'standard functional set' listed above. So that's 13 monads * 13 monads * 14 functions. That's a lot of extension methods. So there's T4 template that generates 'monad transformers' that allows for nested monads.

This is super powerful, and means that most of the time you can leave your Option<T> or any of the monads in this library wrapped. You rarely need to extract the value. Mostly you only need to extract the value to pass to the BCL or Third-party libraries. Even then you could keep them wrapped and use Iter or IterT.

if( arg == null ) throw new ArgumentNullException("arg")

Another horrible side-effect of null is having to bullet-proof every function that take reference arguments. This is truly tedious. Instead use this:

    public void Foo( Some<string> arg )
    {
        string value = arg;
        ...
    }

By wrapping string as Some<string> we get free runtime null checking. Essentially it's impossible (well, almost) for null to propagate through. As you can see (above) the arg variable casts automatically to string value. It's also possible to get at the inner-value like so:

    public void Foo( Some<string> arg )
    {
        string value = arg.Value;
        ...
    }

If you're wondering how it works, well Some<T> is a struct, and has implicit conversion operators that convert a type of T to a type of Some<T>. The constructor of Some<T> ensures that the value of T has a non-null value.

There is also an implicit cast operator from Some<T> to Option<T>. The Some<T> will automatically put the Option<T> into a Some state. It's not possible to go the other way and cast from Option<T> to Some<T>, because the Option<T> could be in a None state which would cause the Some<T> to throw ValueIsNullException. We want to avoid exceptions being thrown, so you must explicitly match to extract the Some value.

There is one weakness to this approach, and that is that if you add a member property or field to a class which is a struct, and if you don't initialise it, then C# is happy to go along with that. This is the reason why you shouldn't normally include reference members inside structs (or if you do, have a strategy for dealing with it).

Some<T> unfortunately falls victim to this, it wraps a reference of type T. Therefore it can't realistically create a useful default. C# also doesn't call the default constructor for a struct in these circumstances. So there's no way to catch the problem early. For example:

    class SomeClass
    {
        public Some<string> SomeValue = "Hello";
        public Some<string> SomeOtherValue;
    }
    
    ...
    
    public void Greet(Some<string> arg)
    {
        Console.WriteLine(arg);
    }
    
    ...
    
    public void App()
    {
        var obj = new SomeClass();
        Greet(obj.SomeValue);
        Greet(obj.SomeOtherValue);
    }

In the example above Greet(obj.SomeOtherValue); will work until arg is used inside of the Greet function. So that puts us back into the null realm. There's nothing (that I'm aware of) that can be done about this. Some<T> will throw a useful SomeNotInitialisedException, which should make life a little easier.

    "Unitialised Some<...>"

So what's the best plan of attack to mitigate this?

  • Don't use Some<T> for class members. That means the class logic might have to deal with null however.
  • Or, always initialise Some<T> class members. Mistakes do happen though.

There's no silver bullet here unfortunately.

NOTE: Since writing this library I have come to the opinion that Some<T> isn't that useful. It's much better to protect 'everything else' using Option<T> and immutable data structures. It doesn't fix the argument null checks unfortunately. But perhaps using a contracts library would be better.

Lack of lambda and expression inference

One really annoying thing about the var type inference in C# is that it can't handle inline lambdas. For example this won't compile, even though it's obvious it's a Func<int,int,int>.

    var add = (int x, int y) => x + y;

There are some good reasons for this, so best not to bitch too much. Instead use the fun function from this library:

    var add = fun( (int x, int y) => x + y );

This will work for Func<..> and Action<..> types of up to seven generic arguments. Action<..> will be converted to Func<..,Unit>. To maintain an Action use the act function instead:

    var log = act( (int x) => Console.WriteLine(x) );

If you pass a Func<..> to act then its return value will be dropped. So Func<R> becomes Action, and Func<T,R> will become Action<T>.

To do the same for Expression<..>, use the expr function:

    var add = expr( (int x, int y) => x + y );

Note, if you're creating a Func or Action that take parameters, you must provide the type:

    // Won't compile
    var add = fun( (x, y) => x + y );

    // Will compile
    var add = fun( (int x, int y) => x + y );

Void isn't a real type

Functional languages have a concept of a type that has one possible value, itself, called Unit. As an example bool has two possible values: true and false. Unit has one possible value, usually represented in functional languages as (). You can imagine that methods that take no arguments, do in fact take one argument of (). Anyway, we can't use the () representation in C#, so LanguageExt now provides unit.

    public Unit Empty()
    {
        return unit;
    }

Unit is the type and unit is the value. It is used throughout the LanguageExt library instead of void. The primary reason is that if you want to program functionally then all functions should return a value and void is a type with zero possible values - and that's the type-theory reason why void is a pain in the arse in C#. This can help a lot with LINQ expressions.

Mutable lists and dictionaries

With the new 'get only' property syntax with C# 6 it's now much easier to create immutable types. Which everyone should do. However there's still going to be a bias towards mutable collections. There's a great library on NuGet called Immutable Collections. Which sits in the System.Collections.Immutable namespace. It brings performant immutable lists, dictionaries, etc. to C#. However, this:

    var list = ImmutableList.Create<string>();

Compared to this:

    var list = new List<string>();

Is annoying. There's clearly going to be a bias toward the shorter, easier to type, better known method of creating lists. In functional languages collections are often baked in (because they're so fundamental), with lightweight and simple syntax for generating and modifying them. So let's have some of that...

Lists

There's support for Cons, which is the functional way of constructing lists:

    var test = Cons(1, Cons(2, Cons(3, Cons(4, Cons(5, empty<int>())))));

    var array = test.ToArray();

    Assert.IsTrue(array[0] == 1);
    Assert.IsTrue(array[1] == 2);
    Assert.IsTrue(array[2] == 3);
    Assert.IsTrue(array[3] == 4);
    Assert.IsTrue(array[4] == 5);

Note, this isn't the strict definition of Cons, but it's a pragmatic implementation that returns an IEnumerable<T>, is lazy, and behaves the same. Functional purists, please don't get too worked up! I am yet to think of a way of implementing a proper type-safe cons (that can also represent trees, etc.) in C#.

Functional languages usually have a shortcut list constructor syntax that makes the Cons approach easier. It usually looks something like this:

    let list = [1;2;3;4;5]

In C# it looks like this:

    var array = new int[] { 1, 2, 3, 4, 5 };
    var list = new List<int> { 1, 2, 3, 4, 5 };

Or worse:

    var list = new List<int>();
    list.Add(1);
    list.Add(2);
    list.Add(3);
    list.Add(4);
    list.Add(5);

So we provide the List function that takes any number of parameters and turns them into a list:

    // Creates a list of five items
     var test = List(1, 2, 3, 4, 5);

This is much closer to the 'functional way'. It also returns a Lst<T> which is an immutable list implementation. So it's now easier to use immutable-lists than the mutable ones. And significantly less typing.

Also Range:

    // Creates a sequence of 1000 integers lazily (starting at 500).
    var list = Range(500,1000);
    
    // Produces: [0, 10, 20, 30, 40]
    var list = Range(0,50,10);
    
    // Produces: ['a,'b','c','d','e']
    var chars = Range('a','e');

Some of the standard set of list functions are available (in LanguageExt.List):

    using static LanguageExt.List;
    ...

    // Generates 10,20,30,40,50
    var input = List(1, 2, 3, 4, 5);
    var output1 = map(input, x => x * 10);

    // Generates 30,40,50
    var output2 = filter(output1, x => x > 20);

    // Generates 120
    var output3 = fold(output2, 0, (x, s) => s + x);

    Assert.IsTrue(output3 == 120);

The above can be written in a fluent style also:

    var res = List(1, 2, 3, 4, 5)
                .Map(x => x * 10)
                .Filter(x => x > 20)
                .Fold(0, (x, s) => s + x);

    Assert.IsTrue(res == 120);

List pattern matching

Here we implement the standard functional pattern for matching on list elements. In our version you must provide at least 2 handlers:

  • One for an empty list
  • One for a non-empty list

However, you can provide up to seven handlers, one for an empty list and six for deconstructing the first six items at the head of the list.

    public int Sum(IEnumerable<int> list) =>
        match( list,
               ()      => 0,
               (x, xs) => x + Sum(xs) );

    public int Product(IEnumerable<int> list) =>
        match( list,
               ()      => 0,
               x       => x,
               (x, xs) => x * Product(xs) );

    public void RecursiveMatchSumTest()
    {
        var list0 = List<int>();
        var list1 = List(10);
        var list5 = List(10,20,30,40,50);
        
        Assert.IsTrue(Sum(list0) == 0);
        Assert.IsTrue(Sum(list1) == 10);
        Assert.IsTrue(Sum(list5) == 150);
    }

    public void RecursiveMatchProductTest()
    {
        var list0 = List<int>();
        var list1 = List(10);
        var list5 = List(10, 20, 30, 40, 50);

        Assert.IsTrue(Product(list0) == 0);
        Assert.IsTrue(Product(list1) == 10);
        Assert.IsTrue(Product(list5) == 12000000);
    }

Those patterns should be very familiar to anyone who's ventured into the functional world. For those that haven't, the (x,xs) convention might seem odd. x is the item at the head of the list - list.First() in LINQ world. xs (many X-es) is the tail of the list - list.Skip(1) in LINQ. This recursive pattern of working on the head of the list until the list runs out is pretty much how loops are done in the functional world.

Be wary of recursive processing however. C# will happily blow up the stack after a few thousand iterations.

Functional programming doesn't really do design patterns, but if anything is a design pattern it's the use of fold. If you put a bit of thought into it, you will realise that recursive processes all tend to follow a very similar pattern.

The two recursive examples above for calculating the sum and product of a sequence of numbers can be written:

    // Sum
    var total = fold(list, 0, (s,x) => s + x);
    
    // Product
    var total = reduce(list, (s,x) => s * x);

reduce is fold but instead of providing an initial state value, it uses the first item in the sequence. Therefore you don't get an initial multiply by zero (unless the first item is zero!). Internally fold, foldBack and reduce use an iterative loop rather than a recursive one; so no stack blowing problems!

Maps

We also support dictionaries. Again the word Dictionary is such a pain to type, especially when there's a perfectly valid alternative used in the functional world: map.

To create an immutable map, you no longer have to type:

    var dict = ImmutableDictionary.Create<string,int>();

Instead you can use:

    var dict = Map<string,int>();

Map<K,V> is an implementation of an AVL Tree (self balancing binary tree). This allows us to extend the standard IDictionary set of functions to include things like findRange.

Also you can pass in a list of tuples or key-value pairs:

    var people = Map( Tuple(1, "Rod"),
                      Tuple(2, "Jane"),
                      Tuple(3, "Freddy") );

To read an item call:

    Option<string> result = find(people, 1);

This allows for branching based on whether the item is in the map or not:

    // Find the item, do some processing on it and return.
    var res = match( find(people, 100),
                     Some: v  => "Hello " + v,
                     None: () => "failed" );
                   
    // Find the item and return it.  If it's not there, return "failed"
    var res = find(people, 100).IfNone("failed");                   
    
    // Find the item and return it.  If it's not there, return "failed"
    var res = ifNone( find(people, 100), "failed" );

Because checking for the existence of something in a dictionary (find), and then matching on its result is very common, there is a more convenient match override:

    // Find the item, do some processing on it and return.
    var res = match( people, 1,
                     Some: v  => "Hello " + v,
                     None: () => "failed" );

To set an item call:

    var newThings = setItem(people, 1, "Zippy");

Obviously because it's an immutable structure, calling add, tryAdd, addOrUpdate, addRange, tryAddRange, addOrUpdateRange, remove, setItem, trySetItem, setItems or trySetItems... will generate a new Map<K,V>. It's quite cunning though, and it only replaces the items that need to be replaced and returns a new map with the new items and shared old items. This massively reduces the memory allocation burden

By holding onto a reference to the Map before and after calling add you essentially have a perfect timeline history of the changes. But be wary that if what you're holding in the Map is mutable and you change your mutable items, then the old Map and the new Map will change.

So only store immutable items in a Map, or leave them alone if they're mutable.

Map transformers

There are additional transformer functions for dealing with 'wrapped' maps (i.e. Map<int, Map<int, string>>). We only cover a limited set of the full set of Map functions at the moment. You can wrap Map up to 4 levels deep and still call things like Fold and Filter. There's interesting variants of Filter and Map called FilterRemoveT and MapRemoveT, where if a filter or map operation leaves any keys at any level with an empty Map then it will auto-remove them.

    Map<int,Map<int,Map<int, Map<int, string>>>> wrapped = Map.create<int,Map<int,Map<int,Map<int,string>>();
    
    wrapped = wrapped.AddOrUpdate(1,2,3,4,"Paul");
    wrapped = wrapped.SetItemT(1,2,3,4,"Louth");
    var name = wrapped.Find(1,2,3,4);               // "Louth"

The Map transformer functions:

Note, there are only fluent versions of the transformer functions.

  • Find
  • AddOrUpdate
  • Remove
  • MapRemoveT - maps each level, checks if the map is empty, in which case it removes it
  • MapT
  • FilterT
  • `FilterRemoveT`` - filters each level, checks if the map is empty, in which case it removes it
  • Exists
  • ForAll
  • SetItemT
  • TrySetItemT
  • FoldT
  • more coming...

The awful out parameter

This has to be one of the most awful patterns in C#:

    int result;
    if( Int32.TryParse(value, out result) )
    {
        ...
    }
    else
    {
        ...
    }

There's all kinds of wrong there. Essentially the function needs to return two pieces of information:

  • Whether the parse was a success or not
  • The successfully parsed value

This is a common theme throughout the .NET framework. For example on IDictionary.TryGetValue

    int value;
    if( dict.TryGetValue("thing", out value) )
    {
       ...
    }
    else
    {
       ...
    }

So to solve it we now have methods that instead of returning bool, return Option<T>. If the operation fails it returns None. If it succeeds it returns Some(the value) which can then be matched. Here's some usage examples:

    
    // Attempts to parse the value, uses 0 if it can't
    int res = parseInt("123").IfNone(0);

    // Attempts to parse the value, uses 0 if it can't
    int res = ifNone(parseInt("123"), 0);

    // Attempts to parse the value, doubles it if can, returns 0 otherwise
    int res = parseInt("123").Match(
                  Some: x => x * 2,
                 None: () => 0
              );

    // Attempts to parse the value, doubles it if can, returns 0 otherwise
    int res = match( parseInt("123"),
                     Some: x => x * 2,
                     None: () => 0 );

out method variants

  • IDictionary<K, V>.TryGetValue
  • IReadOnlyDictionary<K, V>.TryGetValue
  • IImmutableDictionary<K, V>.TryGetValue
  • IImmutableSet<K, V>.TryGetValue
  • Int32.TryParse becomes parseInt
  • Int64.TryParse becomes parseLong
  • Int16.TryParse becomes parseShort
  • Char.TryParse becomes parseChar
  • Byte.TryParse becomes parseByte
  • UInt64.TryParse becomes parseULong
  • UInt32.TryParse becomes parseUInt
  • UInt16.TryParse becomes parseUShort
  • float.TryParse becomes parseFloat
  • double.TryParse becomes parseDouble
  • decimal.TryParse becomes parseDecimal
  • bool.TryParse becomes parseBool
  • Guid.TryParse becomes parseGuid
  • DateTime.TryParse becomes parseDateTime

any others you think should be included, please get in touch

'Erlang like' concurrency

Docs

Another issue with working with C# is that no matter how much of this library you take on-board, you will always end up bumping into mutable state or side-effecting systems. A way around that is to package up the side-effects into atoms of functional computation that are attached to the mutable state (in whatever form it may take).

The Actor model + functional message handling expressions are the perfect programming model for that. Concurrent programming in C# isn't a huge amount of fun. Yes the TPL gets you lots of stuff for free, but it doesn't magically protect you from race conditions or accessing shared state, and definitely doesn't help with accessing shared external state like a database.

This does.

Getting started

Make sure you have the LanguageExt.Process DLL included in your project. If you're using F# then you will also need to include LanguageExt.Process.FSharp.

In C# you should be using static LanguageExt.Process, if you're not using C# 6, just prefix all functions in the examples below with Process.

In F# you should:

open LanguageExt.ProcessFs

What's the Actor model?

  • An actor is a single threaded process
  • It has its own blob of state that only it can see and update
  • It has a message queue (inbox)
  • It processes the messages one at a time (single threaded remember)
  • When a message comes in, it can change its state; when the next message arrives it gets that modified state.
  • It has a parent Actor
  • It can spawn child Actors
  • It can tell messages to other Actors
  • It can ask for replies from other Actors
  • They're very lightweight, you can create 10,000s of them no problem

So you have a little bundle of self contained computation, attached to a blob of private state, that can get messages telling it to do stuff with its private state. Sounds like OO right? Well, it is, but as Alan Kay envisioned it. The slight difference with this is that it enforces execution order, and therefore there's no shared state access, and no race conditions (within the actor).

Distributed

The messages being sent to actors can also travel between machines, so now we have distributed processes too. This is how to send a message from one process to another on the same machine using LanguageExt.Process:

    tell(processId, "Hello, World!");

Now this is how to send a message from one process to another on a different machine:

    tell(processId, "Hello, World!");

It's the same. Decoupled, thread-safe, without race conditions. What's not to like?

How?

Sometimes this stuff is just easier by example, so here's a quick example, it spawns three processes, one logging process, one that sends a 'ping' message and one that sends a 'pong' message. They schedule the delivery of messages every 100 ms. The logger is simply: Console.WriteLine:

    // Log process
    var logger = spawn<string>("logger", Console.WriteLine);

    // Ping process
    ping = spawn<string>("ping", msg =>
    {
        tell(logger, msg);
        tell(pong, "ping", TimeSpan.FromMilliseconds(100));
    });

    // Pong process
    pong = spawn<string>("pong", msg =>
    {
        tell(logger, msg);
        tell(ping, "pong", TimeSpan.FromMilliseconds(100));
    });

    // Trigger
    tell(pong, "start");

Purely functional programming without the actor model at some point needs to deal with the world, and therefore needs statefullness. So you end up with imperative semantics in your functional expressions (unless you use Haskell).

Now you could go the Haskell route, but I think there's something quite perfect about having a bag of state that you run expressions on as messages come in. Essentially it's a fold over a stream.

There are lots of Actor systems out there, so what makes this any different? Obviously I wanted to create some familiarity, so the differences aren't huge, but they exist. The things that I felt I was missing from other Actor systems was that they didn't seem to acknowledge anything outside of their system. Now I know that the Actor model is supposed to be a self contained thing, and that's where its power lies, but in the real world you often need to get information out of it and you need to interact with existing code: declaring another class to receive a message was getting a little tedious. So what I've done is:

  • Remove the need to declare new classes for processes (actors)
  • Added a publish system to the processes
  • Made process discovery simple
  • Made a 'functional first' API

Functions if you want them

If your process is stateless then you just provide an Action<TMsg> to handle the messages, if your process is stateful then you provide a Func<TState> setup function, and a Func<TState,TMsg, TState> to handle the messages (any seasoned functional programmer will notice that is the signature of a fold). This makes it easy to create new processes and reduces the cognitive overload of having loads of classes for what should be small packets of computation.

You still need to create classes for messages and the like, that's unavoidable (Use F# to create a 'messages' project, it's much quicker and easier). But also, it's desirable, because it's the messages that define the interface and the interaction, not the processing class.

Creating something to log string messages to the console is as easy as:

    ProcessId log = spawn<string>("logger", Console.WriteLine);

    tell(log, "Hello, World");

Or if you want a stateful, thread-safe cache:

static class Cache
{
    enum Tag
    {
        Add,
        Remove,
        Get,
        Flush
    }

    class Msg
    {
        public Tag Tag;
        public string Key;
        public ExpiringValue Value;
    }

    class ExpiringValue
    {
        public DateTime Expiry;
        public string Value;
    }

    public static Unit Add(ProcessId pid, string key, string value) =>
        tell(pid, new Msg { Tag = Tag.Add, Key = key, Value = new ExpiringValue { Value = value, Expiry = DateTime.UtcNow.AddMinutes(1) }});

    public static Unit Remove(ProcessId pid, string key) =>
        tell(pid, new Msg { Tag = Tag.Remove, Key = key });

    public static string Get(ProcessId pid, string key) =>
        ask<string>(pid, new Msg { Tag = Tag.Get, Key = key });

    public static Unit Flush(ProcessId pid) =>
        tell(pid, new Msg { Tag = Tag.Flush });

    public static ProcessId Spawn(ProcessName name) =>
        // Argument 1 is the name of the process
        // Argument 2 is the setup function: returns a new empty cache (Map)
        // Argument 3 checks the message type and updates the state except when
        //            it's a 'Get' in which case it Finds the cache item and if
        //            it exists, calls 'reply', and then returns the state 
        //            untouched.
        spawn<Map<string, ExpiringValue>, Msg>(
            name,
            () => Map<string, ExpiringValue>(),
            (state, msg) => 
                match(msg.Tag,
                    with(Tag.Add,    _ => state.AddOrUpdate(msg.Key, msg.Value)),
                    with(Tag.Remove, _ => state.Remove(msg.Key)),
                    with(Tag.Get,    _ => state.Find(msg.Key).Map(v => v.Value).IfSome(reply).Return(state)),
                    with(Tag.Flush,  _ => state.Filter(s => s.Expiry < DateTime.UtcNow))));
}

The ProcessId is just a wrapped string path, so you can serialise it and pass it around, then anything can find and communicate with your cache:

    var pid = Cache.Spawn("my-cache");
    
    // Add a new item to the cache
    Cache.Add(pid, "test", "hello, world");
    
    // Get an item from the cache
    var thing = Cache.Get(pid, "test");

    // Remove an item from the cache
    Cache.Remove(pid, "test");

Periodically you will probably want to flush the cache contents. Just fire up another process, they're basically free (and by using functions rather than classes, very easy to put into little worker modules):

    public void SpawnCacheFlush(ProcessId cache)
    {
        // Spawns a process that tells the cache process to flush, and then sends
        // itself a message in 10 minutes which causes it to run again.
        
        var flush = spawn<Unit>(
            "cache-flush", _ =>
            {
                Cache.Flush(cache);
                tellSelf(unit, TimeSpan.FromMinutes(10));
            });

        // Start the process running
        tell(flush, unit); 
    }

Classes if you want them

For those that actually prefer the class based approach - or would at least prefer the class based approach for the larger/more-complex processes then there is an interface proxy system. The previous Cache example where there's quite bit of boiler-plate because of C#'s lack of discriminated unions and pattern-matching could be implemented thus:

    public interface ICache
    {
        void Add(string key, string value);
        void Remove(string key);
        string Get(string key);
        void Flush();
    }

    public class Cache : ICache
    {
        Map<string, ExpiringValue> state = Map.empty<string, ExpiringValue>();
        
        public void Add(string key, string value, DateTime expires)
        {
            state = state.Add(key, new ExpiringValue(value, expires));
        }
        
        public void Remove(string key)
        {
            state = state.Remove(key);
        }
        
        public string Get(string key)
        {
            return state[key];
        }
        
        public void Flush()
        {
            state = state.Filter(s => s.Expiry < DateTime.UtcNow);
        }
    }

Use it like so:

    // Spawn the Cache process with a state-type of Cache - it accepts the ProxyMsg
    // type for messages which are auto-unpacked and used to invoke methods on the
    // Cache state object.
    ProcessId pid = spawn<Cache>("cache");
    
    // Generate an ICache proxy for calling the methods on Cache.  The proxy function
    // maps the interface onto tell and ask calls, and packs up the method dispatch
    // requests into ProxyMsgs.  It also does a type-check to make sure the Process
    // actually has a state-type of ICache.
    ICache cache = proxy<ICache>(pid);

    // Call the ICache.Add method.  This is translated into a Process.tell 
    cache.Add("test", "hello, world", DateTime.UtcNow.AddMinutes(10));
    
    // Get an item from the cache.  This is translated into a Process.ask
    var thing = cache.Get("test");

    // Remove an item from the cache
    cache.Remove("test");

You could continue to use a stand-alone flush process, but it would need to use the proxy to communicate:

    var flush = spawn<Unit>(
        "cache-flush", _ =>
        {
            proxy<ICache>(pid).Flush();
            tellSelf(unit, TimeSpan.FromMinutes(10));
        });

The proxy can be built from anywhere, the Process system will auto-generate a concrete implementation for the interface that will dispatch to the Process specified. It also type checks your interface against what the actual Process is running adding an extra bit of type-safety to the procedure.

If you only need to work with the Process locally, then you can short-cut and go straight to the proxy:

    ICache cache = spawn<ICache>("cache", () => new Cache());

With the proxy approach we are back in the imperative world. But in some circumstances it is more valuable. If you imagine that each method on ICache is actually an inbox handler, you'll realise we still have the protection of single-threaded access and so the mutable nature of the Process state isn't the concern it would be if it was a regular class.

As you can see that's a pretty powerful technique. Remember the process could be running on another machine, and as long as the messages serialise you can talk to them by process ID or via proxy.

Publish system

Most other actor systems expect you to tell all messages directly to other actors. If you want a pub-sub model then you're required to create a publisher actor that can take subscription messages from other actors; the publisher actor then manages a 'registry' of subscribers to deliver messages to. It's all a bit bulky and unnecessary.

So with LanguageExt.Process each process manages its own internal subscriber list. If a process needs to announce something it calls:

    // Publish a message for anyone listening
    publish(msg);

Another process can subscribe to that by calling:

    subscribe(processId);

(The subscriber can do this in its setup phase, and the process system will auto-unsub when the process dies, and auto-resub when it restarts)

This means the messages that are published by one process can be consumed by any number of others (via their inbox in the normal way).

However, sometimes you want to jump outside of that system. For example, if your code is outside of the process system, it can get an IObservable stream instead:

    var sub = observe<Thing>(processId).Subscribe(msg => ...);

A good example of this is the 'Dead Letters' process, it gets all the messages that failed for one reason or another (serialisation problems, the process doesn't exist, the process crashed, etc.). All it does is call publish(msg), which allows you to subscribe to it for logging purposes. This is how it's defined:

    var deadLetters = spawn<DeadLetter>("dead-letters",publish);

That's it! For a key piece of infrastructure. So it's then possible to easily listen and log issues, or hook it up to a process that persists the dead letter messages.

'Discoverability'

Being able to find other Processes in a cluster (or within the same AppDomain) and dispatch or route to them is essential. There's a supervision hierarchy, where you have a root process, then a child user process under which you create your processes, and in turn they create child processes creating a tree structure with which you can use to route messages locally.

There's also system process under root that handles stuff like dead-letters and various other housekeeping tasks.

    /root/user/...
    /root/system/dead-letters
    etc.

When you create a Redis cluster connection the second argument is the name of the node in the 'cluster' (i.e. the name of the app/service/website, whatever it is). The third argument is the role of the node in the cluster (see Role.Broadcast, Role.LeastBusy, Role.Random, Role.RoundRobin, Role.First - for cluster dispatch methods). There is a static property Process.ClusterNodes that allows you to interrogate the nodes are online and what their role is.

    RedisCluster.register();
    ProcessConfig.initialise("sys", "web-front-end", "web-front-end-1", "localhost", "0");
  • "sys" is the 'system name', but easier to think of it as the name of the cluster as a whole. That means you can call it with a different value and point it at another Redis db for multiple clusters. But for now it's easier to call it sys and leave it.
  • "web-front-end" is the role, you can have multiple nodes in a role and the role based dispatchers allow you to implement high-availability and load balancing strategies.
  • "web-front-end-1" is the name of this node, and should be unique in the cluster
  • "localhost" is the Redis connection (can be comma separated for multiple Redis nodes)
  • "0" is the Redis catalogue to use ("0" - "15")

Then instead of having root as the top level Process in your hierarchy, you have my-stuff:

    /web-front-end-1/user/...
    /web-front-end-1/system/dead-letters

Therefore you know where things are, and what they're called, and they're easily addressable. You can just 'tell' the address:

    tell("/web-front-end-1/user/hello", "Hello!");

Or you can use the ProcessId API to build the path:

   ProcessId a = "/web-front-end-1/user/hello";
   ProcessId b = ProcessId.Top["web-front-end-1"]["user"]["hello"];
   // a == b

Even that isn't great if you don't know what the name of the 'app' that is running a Process. So processes can register by a single name, that goes into a 'shared namespace'. It's a kind of DNS for processes:

    /disp/reg/<name>

To register:

    register(myProcessId, "hello-world");

This goes in:

    /disp/reg/hello-world

Your process now has two addresses, the /web-front-end-1/user/hello-world address and the /disp/reg/hello-world address that anyone can find by calling find("hello-world"). This makes it very simple to bootstrap processes and get messages to them even if you don't know what system is actually dealing with it:

    tell(find("hello-world"), "Hi!");

Along with routers, dispatchers and roles the ability to find, route and dispatch to other nodes in the cluster is trivial. For a full discussion on routing, roles and dispatchers see here

Persistence

There is an ICluster interface that you can use the implement your own persistence layer. However out of the box there is persistence to Redis (using LanguageExt.Process.Redis).

You can optionally persist:

  • Inbox messages
  • Process state

Here's an example of persisting the inbox:

    var pid = spawn<string>("redis-inbox-sample", Inbox, ProcessFlags.PersistInbox);

Here's an example of persisting the state:

    var pid = spawn<string>("redis-inbox-sample", Inbox, ProcessFlags.PersistState);

Here's an example of persisting both:

    var pid = spawn<string>("redis-inbox-sample", Inbox, ProcessFlags.PersistAll);

Process system documentation

The rest

I haven't had time to document everything, so here's a quick list of what was missed:

Type or function Description
TryOption<T> The same as Option<T> except it also handles exceptions. It has a third state called Fail.
Either<Left,Right> Like Option<T>, however the None in Option<T> is called Left in Either, and Some is called Right. Just remember: Right is right, Left is wrong. Both Right and Left can hold values. And they can be different types. See the OptionEitherConfigSample for a demo. Supports all the same functionality as Option<T>.
SomeUnsafe(), RightUnsafe(), LeftUnsafe() These methods accept that sometimes null is a valid result, but you still want an option of saying None. They allow null to propagate through, and it removes the null checks from the return value of match
Set<T>() Creates a Set<T>, an immutable set (AVL tree).
Stack<T>() Creates a Stck<T>, an immutable stack
Queue<T>() Creates a Que<T>, an immutable queue
freeze<T>(list) Converts an IEnumerable<T> to an Lst
memo<T>(fn) Caches a function's result the first time it's called
memo<T,R>(fn) Caches a result of a function once for each unique parameter passed to it
ignore Takes one argument which it ignores and returns unit instead.
Nullable<T>.ToOption() Converts a Nullable<T> to an Option<T>
raise(exception) Throws the exception passed as an argument. Useful in lambda's where a return value is needed.
failwith(message) Throws an Exception with the message provided. Useful in lambda's where a return value is needed.
identity<T>() Identity function. Returns the same value it was passed.

Future

There's more to come with this library. Feel free to get in touch with any suggestions.

About

C# functional language extensions and 'Erlang like' concurrency system

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • C# 94.4%
  • JavaScript 4.6%
  • F# 0.7%
  • HTML 0.2%
  • CSS 0.1%
  • Batchfile 0.0%