Colon Scope

Lets talk a little bit about declaring functions.

rx :: [a] -> a -> [a]
rx [] r = [r]
rx (x:xs) r = x : rx xs r

It is not entirely obvious what the various bits are, so let’s use some long names to make it more readable.

myInfix :: [something] -> something -> [something]
myInfix [] atEnd = [ atEnd ]
myInfix (listHead:listTail) atEnd = listHead : ( myInfix listTail atEnd )

Something magical and weird is happening at the double colon ::  We are declaring the type for our method myInfix.  If this were another language, we would think about the type declaration as:  My Function takes a List and an Element, then produces another List.  If that were the case, we would expect to see the following:

• myInfix :: ( [something] , something) -> [something]

But that conception is false.  Instead, we are saying:  My Function takes a List, then produces another (invisible) Function which takes an Element, which in turn produces a List.  So we can conceptually group that as follows:

• ( (myInfix :: [something]) -> something ) -> [something]           – but do not actually group this way, as it won’t compile

Next to consider ::  ’something‘ is not an actual type.  Instead it represents any type that we decide to use with the function.  The fact that ‘something‘ is mentioned in three places does create a constraint.  We must use and expect the same type in each of the three places.  So the following examples work:

• myInfix [1,2] 6    produces [1,2,6]
• myInfix “hi” ‘o’   produces “hio”

But the following fail:

• myInfix [1,2] ‘n’
• myInfix[False,True] 8

Advertisement