Levente Uzonyi wrote:
> I uploaded a new version of the #nextChunk speedup to the Inbox 
> (Multilingual-ul.85) for reviewing. This one lets the TextConverter do 
> the basic part of the chunk reading, making it fully compatible with any 
> encoding.

That's an excellent version. I like speedups that come at no loss of 
generality :-)

Cheers,
   - Andreas