TODO

Code issues (comments by duncan):
* implement data compression for color as well as
  black and white printing.
* bidirectional printing is not fully consistent (It is not clear
  ahead of time whether the next sweep will be color or black, and thus
  when the sweep changes from black to color or vice-versa, the sweep
  direction may not reverse.)
* check out the new information about the buffer on the printer.
  (mentioned on ppa.devel by andrew in early 2000)
* cleaning head codes needed. (or put them into PUP);
  document the codes in  the ppa docs here.
* should multiple color sweeps in a swath (gMaxpass=4)
  alternate between right-to-left and left-to-right ?
* Printing of e.g. tiger.ps (an example included with ghostscript)
  shows lines between swaths.  The Windows driver does not.  This is because
  the four passes of the color head (gMaxPass=4) are all at the same
  vertical position.   This is not the way to e.g. paint a wall using a
  paint roller - instead the successive passes should  be vertically
  staggered  so  their swath boundaries are not all at the same vertical
  position.   The Windows driver does this: the vertical position of pass 2
  is 1/4 of a swath height  below pass 1, etc.  Unfortunately, the design
  decisions in cutswath.c make this not a simple change.   Initial changes 
  would be to somehow decouple the four sweeps, so they could have independent
  left and right values, directions, vertical positions, etc.

* The Windows driver  also has a "normal" mode with gMaxPass=2, and well
  as "econofast" (gMaxPass=1) and high quality (gMaxPass=4).  It
  alternates color passes in one direction with black passes in the
  other direction, at the same vertical position.
  If this structure is adopted, maybe the way the 600dpi resolution of black
  if reduced to 300dpi resolution in dither.c should be rethought.
  (Why not keep 600dpi black resolution ?).

* A FS_black_dither routine is needed in black-only printing, for high-quality
  black-only images.

* Internationalization ("i18n") should be done with GNU gettext and .po files,
 not the current hack.

* add autodetection of device via IEEE 1284 and other methods
* redo debugging support	(done ajv)
* add extra .conf parameters to avoid recompiles (done: ajv & giorgio)
* getopt() rather than current handling	(done:	ajv 19991003)

Distribution and Installation issues

* *BSD compatibility (partially there: ajv 19991021)
* update README and all doco
* update installation instructions for major distributions (caldera, RH, etc)
* Translate man page into other languages
* add autoconf support ? (NetBSD and alpha issues)
* update CREDITS
* Make install target	(done: ajv)
* make syslog configurable; stdout or syslog (done: ajv 19991021)

CALIBRATION
* CALIBRATE doco, and supporting .ps file to generate page

Long Term wants

* write instructions on how to speed up parallel port (Linux 2.2.x and later)
* write hppaconf (KDE)?
* gs native filter rather than take pnm/pbm/ppm files
* scpd to monitor printer status, and provide support for console alerts
  (These are dumb printers, so ...)