
- Character encodings:
   - Euc-KR
   - GB*
     
- Bugfixes:
   - unihtml input doesn't work. Should process html 1-2 times, not 0-1 times.
   - unihtml input doesn't work, when a html entity wraps between two read blocks (which are short).
   - eucjp output does not seem to work at all.

- Processors:
   - tab and spaces (with tab size parameter, and flag whether to process fully or just line beginnings)
   - base64
   - html entities

- The conversion process should be like this:

   Preprocess -> Decode -> Filter -> Unicode
   Unicode -> Encode -> Postprocess
   
   For example, proprocess=base64, decode=iso-2022-jp,
                filter=html,
                encode=latin1, postprocess=base64
   would decode a base64 encoded stream, then convert iso-2022-jp to unicode,
   and converting html entities from it too, then encode it as 8-bit latin1,
   and re-encode with base64.

   Complicated... But html can't necessarily be preprocessed. It must
   be done between the decode and encode step, not before decode.

- Fix a bug
   
   If the compiling process complains about missing koi8r.h or something
   similar, try removing .depend file and reissuing "make" (don't clean).
