                     Future plans and vague ideas


* Implement loading of encodings dynamically at run time.  Request for
  an unknown encoding would trigger loading a mapping file.  Ideally,
  this would also be implemented in the Type1 rasteriser.  Same problem
  with configuration: you need a search path.

* Make a tool for building fonts.dir files, on the model of mkfontdir.
  This should use the OS/2 table to determine which encodings are
  functional.  Maintaining font.dir files is a nightmare.

* Find out whether its possible to do lazy rasterisation under X.

  This would involve switching the scheme for computing font metrics
  from precise information (deriving it from the bitmaps) to
  approximate information (trusting the font metrics).  As some
  programs (notably terminal emulators) are very sensible to this, it
  should not be done systematically.  Two possibilities are:
    * using the XLFD name (adding `[_]' after the encoding comes to 
      mind);
    * adding a new configuration variable; how do you do it in the X
      server?

* Get rich and famous.
