Am Dienstag, 4. Mai 2004 11:16 schrieb George Russell:

Sven Panne wrote:

...

Hmmm, "String -> [Word8]" would be nicer...

My UTF8 encoder is toUTF8 :: String -> String but an obvious alternative would be toUTF8 :: Enum codedChar => String -> [codedChar] and I could implement this quite easily, by globally-exchanging chr with toEnum. It would then be appropriate to SPECIALIZE to types String -> String and String -> [Word8], satisfying both the purists and those who actually want to write the output to a file.

Writing UTF-8 to a file should be done using binary output anyway, since UTF-8 is a sequence of octets. So Word8 would also be the way to go for the "file writers".

...

... and here: "[Word8] -> String" or "[Word8] -> Maybe String

and my UTF8 decoder has type

fromUTF8WE :: Monad m => String -> m String

Errors are reported by "fail". If for example you import Control.Monad.Error that means you have a function returning either an error message or the converted string

fromUTF8WE :: String -> Either String String

I like this "error handling via monads" and use it myself a lot.

Of course for Word8, you would change the type of the decoder to

fromUTF8WE :: (Monad m,Enum codedChar) => [codedChar] -> m String

Incidentally I am *hoping* I shall be able to say that my UTF8 code is LGPL but you know what University administrators are like ...

Wolfgang