- Complete support for Unicode under MS Windows (e.g. use Unicode
  versions of text-display routines, for better support of less common
  charsets)

<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
   Get minimal Mule for Windows working using Ikeyama's patches.  At
   first, rely on his conversion of internal -> external
   locale-specific but very soon (as soon as we get translation
   tables) can switch to using Unicode versions of display funs, which
   will allow many more charsets to be handled and in a more
   consistent fashion.

   i.e. to convert an internal string to an external format, at first
   we use our own knowledge of the Microsoft locale file formats but
   an alternative is to convert to Unicode and use Microsoft's
   convert-Unicode-to-locale encoding functions.  This gains us a
   great deal of generality, since in practice all charset caching
   points can be wrapped into Unicode caching points.

   This requires adding UCS2 support, which I'm doing.  This support
   would let us convert internal -> Unicode, which is exactly what we
   want.

   At first, though, I would do the UCS2 support, but leave the
   existing way of doing things in redisplay.  Meanwhile, I'd go
   through and fix up the places in the code that assume we are
   dealing with unibytes.

   After this, the font problems will be fixed , we should have a
   pretty well working XEmacs + MULE under Windows.  The only real
   other work is the clipboard code, which should be straightforward.