Re: parallel port (was Arrrrrrrgggg!)

[<aah@nofs.navy.mil>]

I am making some statements off the top of my head, so
someone probably has more complete details that will negate
some of my comments.  However, these are things to be aware of
in any case.  I am not 'dissing' the ECP idea, just trying
to be realistic.
  Tom, you cannot shut off processes in a modern computer
for 40 seconds.  I suggested ECP because it can do DMA and
thereby not require constant attention from the host.  However,
typical port drivers, especially under Windows, require an
initial negotiation phase in order to understand what capabilities
both host and 'printer' have, which is why you may need to have
new drivers written (you are not going to run the port in a
purely standard protocol).
  The ECP FIFO is 16 bytes long.  This is why I have been
recommending that you go to a larger external FIFO someplace
if you want to go this route.
  The DMA transfers used to have a limited size, usually 128KB,
in the old PC world.  I am not sure what that size is with the
DMA controller used with the ECP port.  This usually means that
you set the controller up for xxx bytes of transfer and then
watch the status.  When the transfer is complete, you either
return to the calling program or set the DMA controller up for
the next xxx bytes of transfer.  In other words, we do have a count
of the transfer and don't have to rely on clock timing.
  Not all ECP ports are created equal.  Any driver will probably
have to be tweaked from system to system to make things work.
This is especially true for signal levels; the long cable worked
with your memory card but may not work with a typical parallel port
chip.
Arne