The Haskell 1.4 Report states that instances of MonadZero and
MonadPlus should satisfy these laws:
m >> zero = zero
zero >>= m = zero
m ++ zero = m
zero ++ m = m
Now that MonadPlus has been moved out of the Prelude (and has changed
the names of all its operators), I see that no mention of these laws
is in section 10.2 of the Haskell 98 Library Report. I assume this is
inadvertent and that instances of MonadPlus should still satisfy these
laws.
But I have a question: Would it be reasonable to expect any instance
of MonadPlus to satisfy the following law (using `mplus', not ++ now)?
m >>= (\x->f x `mplus` g x)
=
(m >>= f) `mplus` (m >>= g)
It is true for the Maybe and list instances.
Can anyone give me a reasonable instance of MonadPlus which wouldn't
satisfy this but does satisfy the above four laws? Or, is there some
deep reason why this "should" be true?
Thanks.
- Mark
