-
-
Save lanwin/852890 to your computer and use it in GitHub Desktop.
public static class ObjectExtensions | |
{ | |
public static IEnumerable<T> ToMaybe<T>(this T obj) | |
{ | |
return Equals(obj,null) ? Enumerable.Empty<T>() : new[] {obj}; | |
} | |
} |
Ah I see. But in an ideal world you would have ReturnsMaybeAString() which would already return the IEnumerable and would use the methods above.
@lanwin this is perfectly valid, imho. you did the "unit" part (or must is say function?!?) of the monad. the "bind" part is still missing, although it could be easily applied by means of linq - or your Run() method.
A note aside: Enumberable even provides the singleton list for you: Enumerable.Repeat(t, 1);
The Run method is part of RX and enumerates the enumerable and optionally allows you to pass a side effect to it.
RX also have a EnumerableEx.Return(t). Please have a look at the System.CoreEx assembly of RX which provides a lot useful additions to IEnumerable.
I am not very happy with the naming of my extension method. More correct would be ToEnumerable or AsEnumerable. But the framework has both of them already so i would have collisions.
@lanwin nice. one more reason have a look at RX. The Run() seems to me as if it resembles to the "do" of Haskell. Despite, it would finally render the gazillions of custom ForEach() extensions useless ;)
Hmm as i said it is no 100% like foreach. On first it made no sens that there is an Run() without parameters. But that only run thru the enumerable without accessing the result.
They also provide a Do. But that allows you to add a side effect while enumerating the enumerable. But its dose not start the enumeration. Example:
GetList()
.Where(item=="foo")
.Do(item=>Console.WriteLine(item))
.Run();
lanwin is right. Since IEnumerable is lazy you need something which starts the process. Foreach() will do so but breaks the monad. Do() is something which stays in the monad and therefor keeps laziness.
A slight, but noteworthy difference. I don't know RX, so bare with me. I assumed that RX's Run() is the same as you have used it in your comment - which for me is a simple application like a commonly used ForEach-Extension does.
I wasn't aware of that "do" adds side effects without application. This fact actually is a clear indicator that i yet have a long learning path before me.
This way i can avoid unnecessary null checks.