Why does (>>) require ()

[axman6]

I was surprised to see the type of (>>) requires () and not an arbitrary type, I'm wondering if this is intentional, and if so, it should probably come with an explanation, because it doesn't match Monad's (>>),

stacked/src/Control/Monad/Indexed.hs

Line 92 in 82ac628

(>>) :: (Applicative m) => m i j () -> m j k a -> m i k a

[aspiwack]

It is intentional, in that I think that the type of (>>) in Base is a mistake. I think that most everybody agrees with me nowadays since there's a warning for it, namely unused-do-bind.

It forces you to use void or an explicit binding.

Now, I haven't been consistent with this: I've chosen *> and <* to drop the return values silently. Because, to be quite honest, there is a pattern in parser combinators to return values that you don't need because they're free to ignore. And I was worried that forcing the dropped side to be () would make it unnecessarily annoying to port existing parsers to Pup.

So maybe the real answer is: maybe the warning is enough. And I can have the more unsurprising type for >>.

What do you think?