On Thu, Dec 15, 2011 at 9:13 AM, Gregory Crosswhite
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