Discriminated unions, pattern matching and partial applications for C#
Succinc<T> is a small, but growing, .NET library that adds a number of functional features to C#:
"Implicitly" typed lambdas,
The ability to treat void methods as Unit functions,
Replacements for TryParse methods that return an Option<T> (or Maybe<T> , if you prefer), rather than using the out parameter anti-pattern.
"cons" support for IEnumerable<T> (add elements to the head of an enumeration, or split an enumeration into its head element and an enumeration containing the remaining items, all without repeatedly enumerating that enumerable).
Cycle() methods that endlessly repeat an enumeration or array, again without repeatedly enumerating that enumerable).
Replacements for IEnumerable<T> 's XXXOrDefault methods that return an Option<T> (or Maybe<T> , if you prefer), avoiding null and the "did it return a value, or the default?" problem,
And finally, as an experimental feature at this stage, forward pipe support.
The current release of Succinc<T> is 2.1.0, which is available as a nuget package that supports .NET 4+, Windows 8+ appstore apps, Windows Phone 8.1 apps and .NET Core .
This release includes:
"cons" support for IEnumerable<T> . This allows items to be added to the head of an enumeration and for enumerations to be split into the head item and the "tail" enumeration.
Cycle() methods that can endlessly repeat an enumeration without repeatedly enumerating it.
A new experimental feature of forward piping parameters into method calls.
Support for .NET Core.
Succinc<T> provides a set of union types ( Union<T1, T2> , Union<T1, T2, T3> and Union<T1, T2, T3, T4> ) where an instance will hold exactly one value of one of the specified types. In addition, it provides the likes of Option<T> and Maybe<T> that can have the value Some<T> or None .
Succinc<T> uses Option<T> to provide replacements for the .NET basic types' TryParse() methods and Enum.Parse() . In all cases, these are extension methods to string and they return Some<T> on a successful parse and None when the string is not a valid value for that type. No more out parameters! See theOption Parsers guide for more details.
Further Succinc<T> uses Option<T> to provide replacements for the XxxxOrDefault LINQ extension methods on IEnumerable<T> . In all cases, these new extension methods, eg TryFirst<T>() return an option with a value if a match occurred, or None if not.
Succinc<T> can pattern match values, tuples, unions etc in a way similar to F#'s pattern matching features. It uses a fluent (method chaining) syntax to achieve this. Some examples of its abilities:
Succinc<T> offers an alternative approach, taking advantage of the fact that var can be used if the result of a method is assigned to the variable. Using the Func , Action , Transform and Lambda set of methods, the above can be expressed more simply as:
var times = Lambda<double>((p1, p2) => p1 * p2);
For functions, the Lambda methods can be used when the parameters and return type all have the same value, as above. This means the type parameter need only be specified once. Transform can be used when all the parameters are of one type and the return value, another. The Func methods are used when the parameters and/or return type are of different values. For actions, Lambda can also be used when only one type is involved and the Action methods do a similar job to the Func methods.
This is explained in detail in the Succinc<T> typed lambdas guide .
Action / Func conversions
The ToUnitFunc methods supplied by Succinc<T> can be used to cast an Action lambda or method (from 0 to 4 parameters) to a Func delegate that returns unit , allowing void methods to be treated as functions and thus eg used in ternary oprators. In addition, Succinc<T> provides an Ignore method that can be used to explicitly ignore the result of any expression, effectively turning that expression into a void statement.
These methods are explained in detail in the Succinc<T> Action / Func conversions guide .
In many functional languages, collections of data are (can be) treated as linked lists and they have the built in pattern of treating that list as the head (first element) and tail (the rest of the list). This pattern is referred to as "cons". It allows recursive composition and decomposition of the list via a simple syntax, such as:
let newList = newItem :: existingList
let head :: tail = newList
The first line adds newItem to the list, exstingItem , to create a new list, which is assigned to newList .
The second, splits newList into the value held in the first element, and assigns it to head . In addition, the remainder of newList is assigned to tail . So newItem and head have the same value and tail and existingList contain the same elements.
Succinc<T> introduces the same idea to C#, but for all types that implement IEnumerable<T> . Following the immutable-by-default conventions of functional languages, items can be added and removed from the head of an enumeration without affecting the original collection. Also, in keeping with the basic tenet of IEnumerable<T> (that it should only be enumerated once), this is also done in a way that avoids re-enumerating the collection. This is achieved by caching the collection as it is enumerated.
So for an IEnumerable<T> , enumeration , the above lines with Succinc<T> would be:
var newList = enumeration.Cons(newItem);
var consResult = newList.Cons();
var head = consResult.Head;
var tail = consResult.Tail;
For more details, see theCons help page.
Sometimes, with declarative programming, it's useful to be able to treat a set of data as an endlessly repeating set. Haskell's solution to this is the cycle functions, which endlessly repeat a list. Succinc<T> provides Cycle methods too, as either an extension method for IEnumerable<T> , or as a method that takes a list of values as it's parameters.
An example of their use is in solving the "fizzbuzz" problem, as shown below:
var fizzes = Cycle("", "", "Fizz");
var buzzes = Cycle("", "", "", "", "Buzz");
var words = fizzes.Zip(buzzes, (f, b) => f + b);
var numbers = Range(1, 100);
var fizzBuzz = numbers.Zip(words, (n, w) => w == "" ? n.ToString() : w);
In the above code, fizzBuzz is a 100 element enumeration of the form "1", "2", "Fizz", "4", "Buzz", etc. All achieved without using a single if or loop.
For more details, see theCycle help page.
Another feature in many functional languages is the concept of the pipe operator. The idea behind it is that, rather than expressing a series of functions - where each returns a value which is passed to the next function - as:
let result = f4(f3(f2(f1(value))))
The expression can instead be written in an easier to read form:
let result = value |> f1 |> f2 |> f3 |> f4
Succinc<T> offers similar functionality. Custom operators aren't supported in C#, so the syntax isn't as neat as the above, but by using an extension method, something close can be achieved:
var times = Lambda<double>((p1, p2) => p1 * p2);
var times8 = times.Apply(8);
var result = times8(9); // <- result == 720
For more details, see the Forward Pipe Operator page .