oneOf does not seem to do the right thing

[ad-si]

> Foldable.oneOf (Nothing : [Just "something"])
(Just "something")

> NonEmpty.oneOf (Nothing :| [Just "something"])
[Nothing, (Just "something")]

[garyb]

It's a different oneOf - you're not comparing equivalent operations here. In Foldable terms this is

Foldable.oneOf ([Nothing] :| [[Just "something"]])

since the NonEmpty version is using the Alternative instance of the f, not the g. NonEmpty.oneOf is guaranteed to always give you a result, that's not true/possible in the Foldable case :)

There's some discussion on #1

[ad-si]

Well ok, but the function name is very misleading then. It should probably be changed, right?
E.g. how about allOf? ^^

[hdgarrood]

I don't think it's the function name which is misleading, I think this is just a potentially surprising behaviour due to the fact that you're using two different Alternative instances here. Changing the name to allOf wouldn't make sense for NonEmpty Maybe, for instance:

> NonEmpty.allOf ("hi" :| Just "bye")
Just "hi"

In fact, for the majority of Alternative instances, I think allOf would be more misleading than oneOf.

Whenever you're working with the Alternative Array instance, everything will make more sense if you think of Array a as a "nondeterministic computation" which produces a single result of the type a nondeterministically. Of course you have to squint a bit because it's not actually nondeterministic, but from this viewpoint many of the Array instances make more sense, e.g. f <$> x <*> y can be thought of as the nondeterministic computation which calls f with the results of the nondeterministic computations x and y. In the case of NonEmpty Array, the NonEmpty.oneOf function is the function which takes a nondeterministic computation which is guaranteed to be able to produce a result and returns one of its potential results.