[Haskell-cafe] Desugaring of infix operators is (always?) the wrong way round

[Brian Hulley]

Ryan Ingram wrote:
> My comments inlined below...
>
> On 9/25/07, Brian Hulley <brianh at metamilk.com> wrote:
>   
>>   let
>>        shiftLeftByThree = shiftL' 3
>>   in
>>       map shiftLeftByThree  [10, 78, 99, 102]
>>     
>
> let shiftLeftByThree = (`shiftL` 3) in ...
>
>   
Aha! but this is using section syntax which is yet another complication. 
Hypothesis: section syntax would not be needed if the desugaring order 
was reversed.
>> Can anyone think of an example where the current desugaring of infix
>> arguments gives the correct order when the function is used in a postfix
>> application? (apart from commutative functions of course!)
>>     
>
> A couple off the top of my head:
>
> (:) :: a -> [a] -> [a]
>   

Yes that's one that had totally slipped my mind ;-)
> <|> :: MonadPlus m => m a -> m a -> m a
> (how do you define "correct" in this case, anyways?)
>   

I'm not so sure about this one eg:

     first <|> second
     (trychoice second) first

because (trychoice second) encapsulates what to do when its argument 
action fails.
> Even for "shift" I can think of several reasons to want to use it both
> ways; for example, unpacking a bitfield from a Word16:
>
> unpack v = (getM 0 255, getM 8 1, getM 9 31, getM 14 3)
>     where getM = (.&.) . (shiftR v)
>
>   
I suppose with some ops perhaps there is no "most common" way of wanting 
to use them and hence no "one-true-way" argument order for those ops.

Thanks for the examples,
Brian.