[Haskell-cafe] There is no null; Maybe/Option types

[Ertugrul Soeylemez]

wren ng thornton <wren at freegeek.org> wrote:

> But the greatest thing about Maybe is that you don't *have* to write
> code in monadic style. Because Maybe makes explicit the null-pointer
> shenanigans in other languages, you can simply unwrap the Maybe and
> pass around the raw value instead of letting Nothing permeate your
> whole program. Spending all your time in the Maybe monad is just as
> bad as spending all your time in the IO monad.

Unless you're saying that I'm one of the worst Haskell programmers in
the world, you are totally wrong.  Monads are an abstraction for
convenience and beauty in expression, not for encapsulating bad habits.
Particularly there is nothing wrong with writing 90% of your code in
monadic style, which is what I do, often combining three or more monads.

And even if I don't use Maybe in monadic style, I often use combinators
like 'maybe'.  Monadic style to avoid explicit wrapping, and combinators
to avoid explicit unwrapping.

What exactly is wrong with that?  Why would you /not/ use (>=>) to
calculate the fourth root from a monadic square root function?

  fourthRoot :: (Alternative f, Integral i, Monad f) => i -> f i
  fourthRoot = mSqrt >=> mSqrt


> Purity begets clarity!

All Haskell code is pure (in the sense of referential transparency),
including monadic code, even in the IO monad.


Greets,
Ertugrul


-- 
nightmare = unsafePerformIO (getWrongWife >>= sex)
http://ertes.de/