Hello!
On Mon, Dec 27, 2010 at 9:41 AM, Evan Laforge
So it sounds like the consensus is to bite the bullet and try converting to ByteString + attoparsec and see if that helps. Or write attoparsec-text myself, or wait for someone else to do it :)
I might not get around to this real soon, but I'll post my results when (if) I do.
I've uploaded attoparsec-text and attoparsec-text-enumerator to Hackage. I've written those packages late last week and asked for comments from attoparsec and attoparsec-enumerator's maintainers. Although both packages weren't stress tested (actually they were very lightly tested), I'm releasing them so that we don't waste efforts duplicating work =). I'll make an announcement later today if no critical bugs are found. Please test it and try to break it =). Bonus points if we give us some numbers about how it compares to your Parsec 2/3 approach. Links: http://hackage.haskell.org/package/attoparsec-text http://hackage.haskell.org/package/attoparsec-text-enumerator Cheers! =D -- Felipe.