Hacker News new | ask | show | jobs
by lallysingh 4485 days ago
Even if you're not curious about Haskell, have a look at the parser here. Parsing in Haskell feels like cheating.
2 comments
tel 4485 days ago
(And is a really great example motivating both monads and monad transformers, for the curious!)
link
tel 4484 days ago
For the extra curious, here's a little Parser (untested, poor design, but generally the right idea)

    newtype Parser a = Parser { unParser :: MaybeT (State String) a }
      deriving ( Functor, Applicative, Alternative, 
                 Monad, MonadPlus, MonadTrans, 
                 MonadState String )
        -- why write boilerplate when the 
        -- compiler will for you?

    runParser :: Parser a -> String -> Maybe a
    runParser input = flip evalStateT input . runMaybeT 

    -- | Parses a single character if it passes a predicate
    satisfy :: (Char -> Bool) -> Parser Char
    satisfy p = do
      (c:cs) <- get
      guard (p c)
      put cs
      return c
 
    char :: Char -> Parser Char
    char c = satisfy (== c)

    -- | Parses a whole string
    string :: String -> Parser String
    string = mapM char

    -- | Converts a parser to be surrounded by parentheses
    parens :: Parser a -> Parser a
    parens p = char '(' *> p <* char ')'
link
freyrs3 4485 days ago
The shallow DSL in the LLVM monad is really nice, it looks almost like the IR itself but you can abstract over it and compose it with other code. DSLs are often overlooked when talking about monad use cases.
link
tel 4484 days ago
I really need to explore LLVM in more detail at some point.
link
casion 4485 days ago
If you're not curious about haskell, I suspect that parser is going to be very difficult to grok.

At least a basic understanding of haskell (or similar) is going to be required to appreciate it, and at that point you're obviously someone who was curious enough to have the knowledge to understand it to begin with.

link