Re: simple question about using manyTill

[John MacFarlane <jgm-TVLZxgkOlNX2fBVCVOL8/A@public.gmane.org>]

+++ Jeff [May 24 17 08:35 ]:
>   I am new to Haskell and I am trying to write a new reader for Pandoc.
>   There is something strange about manyTill and I'm a bit stuck.
>   I have written a minimal code to present the problem. Suppose I want to
>   write a parser that succeeds if the string ends in three equal signs
>   "===", and return the string preceeding the "===". I have written three
>   versions. One with Parsec, and two with Pandoc:
>   import Text.Parsec.String (Parser)
>   import Text.Parsec (parse, ParseError)
>   import Text.Parsec.Combinator (many1, manyTill)
>   import Text.Parsec.Char (anyChar, string, noneOf)
>   import qualified Text.Pandoc.Builder as B (str)
>   import Text.Pandoc.Builder (Inlines)
>   import Text.Parsec.Prim (try)
>   simpleParse :: Parser a -> String -> Either ParseError a
>   simpleParse p = parse p ""
>   header' :: Parser String
>   header' = manyTill anyChar (string "===")
>   header'' :: Parser [Inlines]
>   header'' = manyTill (B.str <$> (many1 anyChar)) (string "===")
>   header''' :: Parser [Inlines]
>   header''' = manyTill (B.str <$> (many1 (noneOf "="))) (string "===")
>   I thought header' and header'' are the same parser except the type is a
>   bit different.

There's an important difference.  header' will parse one
character at a time (anyChar), each time stopping to see
if the end condition (string "===") is met.

header'' will parse CHUNKS of one or more character at
a time (many1 anyChar), only checking after each chunk
to see if the end condition is met.  Does that explain it?