MR K P SCHUPKE wrote: > >In the general case, it needs to be a bit more complex than that, > > Thats why the functions handled lists not individual characters, > I was assuming that each [Word8] -> [Char] represented a valid > and complete encoding block... IE at the start of each call it > assumes no escapes. All this means is than when reading in chunks > you paste those chunks together before conversion, and you can > only break outside of escapes. This in my opinion is better > behaviour anyway... I don't want some hidden escape state mangling > output, just because some earler code generated invalid output.
Right. Certainly, a stateless interface will handle converting "complete" strings (pathnames, arguments, etc). But, ultimately we will have need of a more general interface. E.g. in the chunked HTTP example which Oleg gave, you would probably want separate decoders for the headers and body, switching between them as you read the stream. You wouldn't want to have to accumulate the entire body as a single byte string just so that you could decode it in one go, and you can't just "push" a decoder onto the stream. -- Glynn Clements <[EMAIL PROTECTED]> _______________________________________________ Haskell-Cafe mailing list [EMAIL PROTECTED] http://www.haskell.org/mailman/listinfo/haskell-cafe
