[Haskell-cafe] Is (==) commutative?

[Alexander Solla]

Depending on the context, it may or may not be wise to distinguish
between undefined and [undefined].  This is a matter of strictness,
laziness, and totality.  If you identify all bottoms as one, you
essentially restrict yourself to (what might as well be, for the
purposes of this discussion) a strict subset of the Haskell language.

If you distinguish between values that "contain" bottom and "are"
bottom, then you need to deal with the semantics of laziness.

Like I said, it is classically valid, so choosing these semantics
means that it will be safe.  But the undecidability of comparing
distinct bottoms means that you have to make a choice.

On 7/24/12, Christian Sternagel <c.sternagel at gmail.com> wrote:
>> It's a classically valid inference, so you're "safe" in that respect,
>> and it is true that evaluating x == y requires traversing x and y, so
>> that if x or y "are" bottom, (x == y) and (y == x) will both be bottom.
> Well, (x == y) could result in bottom, even if neither x nor y are
> bottom, e.g., [undefined] == [undefined]. -cheers chris
>