Misleading Use of Currying in Type Signatures

Question

Misleading Use of Currying in Type Signatures

Opened this issue 13 years ago · 7 comments

Ejhfast commented 13 years ago

Jameshfisher of reddit made a good point:

http://www.reddit.com/r/javascript/comments/pholi/typedjs_sanity_check_your_code/c3pscxo

panesofglass commented 13 years ago

Nice!

Answer 1 · 2012-02-11T10:43:31.000Z

Bear in mind that the right-associative arrow does make sense where currying is actually going on. So Char -> Char -> Char would be the appropriate signature for

//+ char_first :: Char -> Char -> Char
function char_first(c1){
  return function(c2) {
    return c1;
  }
}

(I'm Jameshfisher :)

Answer 2 · 2012-02-11T16:52:48.000Z

Right. But knowing whether a function can be curried (like your example) requires a bit of analysis that I'm not currently doing ;-) Maybe an extra feature later on.

Answer 3 · 2012-02-11T18:49:26.000Z

I can't remember the Haskell equivalent exactly, but I think it would be //+ char_first :: Char Char -> Char. Another option, if the delimiter is necessary, might be //+ char_first :: Char * Char -> Char, a la ML.

Answer 4 · 2012-02-26T10:33:31.000Z

//+ add :: Number, Number -> Number currently works.

Answer 5 · 2012-03-07T18:14:55.000Z

Love the principle of TypedJS but having some problems using it so, with apologies if I'm being a "DA"....

Does currying work? I tried a simple function:

fnCurry = function (a) { return function (b) { return true; } };

and then used

test = TypedJS.addTest("fnCurry :: Number -> Number -> Boolean", fnCurry);
TypedJS.go([test]);

but it fails 300 times with

Expected "boolean" but returned undefined on input: [62,33]

Answer 6 · 2012-03-07T18:53:33.000Z

No, currying doesn't work :(