[Haskell-cafe] Re: Who is afraid of arrows, was Re: ANNOUNCE: Haskell XML Toolbox Version 9.0.0

[Heinrich Apfelmus]

C. McCann wrote:
> Heinrich Apfelmus wrote:
>>> Combined with >>= / >> you have multiple reading direction in the same
>>> expression, as in
>>>
>>> expression      ( c . b . a ) `liftM` a1 >>= a2 >>= a3
>>> reading order     6   5   4            1      2      3
>>
>> That's why I'm usually using  =<<  instead of  >>= .
> 
> Does it bother you that (=<<) is defined to be infixr 1, while (<$>)
> and (<*>) are infixl 4? Or is that just me?
> 
> For instance, I might write the above expression as something like:
> 
> a3 =<< a2 =<< a . b . c <$> a1
 >
> But this still seems awkward, because it mixes different fixities and
> I have to mentally regroup things when reading it. Right associativity
> here does make a certain amount of sense for monads, but
> left-associativity is consistent with plain function application and
> feels more natural to me.

Well, you can't give  (=<<)  left fixity because its type doesn't allow it.

   (a3 =<< a2) =<< a1   -- ill-typed!

So, (=<<)  is modeled after  (:) , not after (<$>) .


Regards,
Heinrich Apfelmus

--
http://apfelmus.nfshost.com