These are the bugs and presumed bugs that I have found in the PostScript
interpreter of the XEROX 4520 printer.

* The _image_ operator produces no output on _showpage_ unless there is also
  some non-raster graphics on the page. (No error message, just goes back to
  "Online Ready"

* Certain _ioerror_ faults cause the interpreter to ignore all further input
  even though the error page has printed out and the display shows "Online
  Ready". This can be easily reproduced by putting insufficient hex data
  in an _image_ stream.

* A "timeout" error occurs after sitting for a few minutes unused. The error
  message prints out even if error reports are disabled.

* Once in "Power Save" mode it will not come out of sleep when data is sent
  to it. Even in PCL mode it takes a long time (a minute or more) but does
  eventually wake up. It should be programmed to initiate wakeup procedures
  as soon as the first byte is received.

* The following binary characters, when embedded in a data stream for use
  with the _image_, _readstring_, _filter_, or other binary operators, are
  trapped and executed as instructions:
  03h ^C, executes _interrupt_
  04h ^D, treated as EOD, after which the rest of the binary data stream is
    scanned for a PostScript command, forcing _limitcheck_
  11h,13h,14h ^Q,^S,^T each cause the scanner to "backspace", overreading
    into the following PostScript command.

  On page 74 of the Red Book, 2nd edition, "Control Codes are device dependent
  and not part of the PostScript language." Yet it is not a communications
  protocol intercepting these characters, but the PostScript interpreter
  itself. If it is desired to trap ^C to allow unscheduled termination of
  a print job, this should be disabled while scanning raw binary data.