To quote Ross Paterson's proposals:

instance Alternative [] where
...
some [] = []
   some (x:xs) = repeat (repeat x)

   many [] = [[]]
   many (x:xs) = repeat (repeat x)

Isn't this instance conflating the ZipList notion and the nondeterministic list notion?

• some v = (:) <$> v <*> many v
• many v = some v <|> pure []

Is there a motivation for writing the second law like this and not like

many v = pure [] <|> some v

other than "parsers should try longest match first"? Because apart from that, I find the version with flipped arguments to <|> more natural (base case first). Incidentally, it would lead to terminating definitions of 'some' and 'many' for lists:

ghci> take 5 . map (take 5) $ some [1,2]

[[1],[1,1],[1,1,1],[1,1,1,1],[1,1,1,1,1]]

ghci> take 5 . map (take 5) $ many [1,2]

[[],[1],[1,1],[1,1,1],[1,1,1,1]]

Sebastian