Reputation: 131
ggt_euklid :: Nat1 -> (Nat1 -> Nat1)
I am trying to learn partial application, I know that in this case, if the parentheses would be left out, I would get the same result, but I do not know how this signature should be evaluated.
As far as I have understood, parentheses signify that it is a function? Would that not imply that ggt_euklid takes a value Nat1 and returns a function?
Below is the complete function:
ggt_euklid x y
| x == y = x
|x>y =ggt_euklid(x-y) y
|x<y =ggt_euklid x (y-x)
Upvotes: 0
Views: 78
Reputation: 2076
Would that not imply that ggt_euklid takes a value Nat1 and returns a function?
No, it still imply that ggt_euklid
takes one argument of type Nat1
and return a function of type Nat1->Nat1
, even though, the parentheses be left out.
The Arrow ->
always be right-associativity (when no parentheses), i.e.:
Nat1 -> Nat1 -> Nat1
is equivalent to
Nat1 -> (Nat1 -> Nat1)
which is corresponding to the function application always be left-associativity. (when no parentheses) for example:
ggt_euklid 1 2
is equivalent to
(ggt_euklid 1) 2
Here
(ggt_euklid 1) ~ Nat1 -> Nat1
and
(ggt_euklid 1) 2 ~ Nat1
So, no matter whether one or two arguments apply to ggt_euklid
, it always return a function of type Nat1 -> Nat1
firstly, if second argument is provided, it applies second argument to the returned function.
Upvotes: 2
Reputation: 92117
You have understood the type signature correctly: it takes one argument and returns a function. That's how "multi-argument" functions in Haskell work: through currying. You can see this in action by trying this equivalent implementation:
ggt_euklid :: Nat1 -> (Nat1 -> Nat1)
ggt_euklid x = \y -> result
where result | x == y = x
| x > y = ggt_euklid (x-y) y
| x < y = ggt_euklid x (y-x)
Here I've introduced this fairly pointless result
variable as a thing to use your pattern guards on, but the idea is the same.
Upvotes: 4