Fix: #1690 - Infix typed holes are now filled using infix notation

[OliverMadine]

fix: #1690

In follow up to my response to @July541's PR (#1694), I have written a different solution to cover the cases of filling holes with infix operators.

applying the "replace _ with (+)" code action to

test :: Int -> Int -> Int
test a1 a2 = a1 `_` a2

now results in

test :: Int -> Int -> Int
test a1 a2 = a1 + a2

rather than

test :: Int -> Int -> Int
test a1 a2 = a1 `(+)` a2

Note: This solution depends on processHoleSuggestions returning infix functions in their prefix form (e.g. (+)) to determine if a function infix/prefix. I imagine there might be a nicer way of doing this.

Edit: In some commit messages I have written "postfix" by mistake instead of "prefix"

[jneira]

LGTM, thanks for the fix, tests cases looks great, many thanks!

[jneira]

This solution depends on processHoleSuggestions returning infix functions in their postfix form (e.g. (+)) to determine if a function infix/postfix. I imagine there might be a nicer way of doing this.

The unique caveat i can think of is somebody is using already postfix notation and we turn it in infix one, is there some test case to check we will not do it?

[OliverMadine]

This solution depends on processHoleSuggestions returning infix functions in their postfix form (e.g. (+)) to determine if a function infix/postfix. I imagine there might be a nicer way of doing this.

The unique caveat i can think of is somebody is using already postfix notation and we turn it in infix one, is there some test case to check we will not do it?

Do you mean that we should check that infix form is not used when prefix is required?

So we should make sure

foo :: Int -> Int -> Int
foo = _ a1 a2

is filled to

foo :: Int -> Int -> Int 
foo a1 a2 = (+) a1 a2

?

I've added this as a test case.

[jneira]

yeah, that was my idea, many thanks