>>>>> "David" == David Brownlee <abs@NetBSD.org> writes:

    >> The appended patch solves the problem completely for me (and
    >> about 40 tcp capable print spoolers of our dept.). The
    >> important part is the 'select' call itself (waiting for a sent
    >> back message and timeout), which does not end with the
    >> 'timeout' normally. [...]

    David> Out of curiosity how long does that select() need to be?
    David> If you tried a piece of postscript which takes a long time
    David> to execute, then generates something on stdout at the end
    David> does it still work?

    David> Would you be willing to add a count of how long has been
    David> waited to the syslog() and record the results?

    David> Sample postscript which tends to take quite a while to
    David> process attached :)

Here we go:

- For your postscript sample (raytrace.ps) the printer takes about 38
  seconds to mechanically start the print job after begin of receiving
  data.

- The select call starts about 3 seconds before start of mechanical
  printer activity. Its end/return coincides with the start of
  mechanical printer activity.

More I should test?

Markus.