Hi guys, I've just started the lzw filter implementation :) It has an "EarlyChange" param, and I was trying to guess what the param's hash maps.
I think that the best behaviour is to map pdf_char_t* to pdf_char_t* and let the filter decode the param content, as otherwise it would be quite error prone for the library user as she'd be tempted to store the param value in an incorrect type. This would ease also the implementation of the PDF parser, abstracting it from filters internals. What do you think? I will start using a pdf_char_t* anyway and change it later if you prefer :) JP
