[Rxtx] loopback puzzle

[Andy Dalton]

One interim work-around (read 'hack') might be to do this externally
to rxtx.  There has been some talk about tying into hotplug
functionality, but it might be easier to listen for file system
modification events.  You could watch /dev, and if an entry is created
that matches what you're looking for as a serial port, call
CommPortIdentifier.addPortName(...).  If that worked, it'd only solve
half the problem -- you still wouldn't be able to remove ports.

I know Mac OSX shares the FreeBSD 'kqueue' API, which seems to be
something similar to Linux's 'inotify' (or vice versa).  There are
Java APIs that wrap 'inotify' (see http://jnotify.sourceforge.net/),
but I don't know if there are any for 'kqueue'.

Andy

On 9/11/06, Dr. Stephen Judd <sjudd at seas.upenn.edu> wrote:
> Your comment about the kernel probably indicates you are discussing
> Linux.
> I'm running on Mac OSX 10.4.7, which traditionally has a big emphasis on
> plug and play. Can I get those ports to appear on the Mac? How?
> Do I have to use something other than RXTX to get there today?
> sj
>
> On 2006 Sep 11, at 7:29 PM, Trent Jarvi wrote:
>
> >
> > Hi Stephen
> >
> > The kernel probably already does everything needed.  Eventually, we
> > will
> > be receiving those events and sending events to applications that a
> > device
> > was plugged in. Depending upon the OS, either the kernel events are
> > not
> > detectable or rxtx ignores them at this time.
> >
> > Right now the commports are enumerated when RXTXCommDriver's static
> > initialize is called.  Thats usually the first
> > 'CommPortIdentifier.getPortIdentifiers();' thats called. This is in
> > RXTXCommDriver.java.  A small enhancement that many folks would
> > like would
> > be to allow rescanning of ports.  Restarting the application after
> > changing hardware is the usual path.  Also consider that we do not
> > currently make special efforts to deal with hardware vanishing.
> >
> > Whatever error you see after hardware is removed was not thought of
> > as a
> > flow control path for UARTs being ripped off the motherboard.  It just
> > wasn't something that happened.
> >
> > On Mon, 11 Sep 2006, Dr. Stephen Judd wrote:
> >
> >> AH!  great.
> >> So my mental model was reasonable, but ahead of its time.
> >> Now how do I adjust my mental image? Does rebooting the machine
> >> change anything? How do I get to those new ports?!
> >> sj
> >>
> >> On 2006 Sep 11, at 7:00 PM, Trent Jarvi wrote:
> >>
> >>>
> >>> Hi Stephen
> >>>
> >>> The CommAPI 'specification' was written before ports could vanish
> >>> and
> >>> appear.  There isn't a means of catching 'hotplug' events yet.  For
> >>> all
> >>> practical purposes, the hardware should not change while using
> >>> CommAPI
> >>> 2.0.  This shortfall has been discussed as a possible addition to
> >>> future
> >>> 'standards' which rxtx will support.  For now, you have to adjust
> >>> your
> >>> mental image :)
> >>>
> >>> On Mon, 11 Sep 2006, Dr. Stephen Judd wrote:
> >>>
> >>>> Many thanks to Dr. Lyon, Joachim Buechse, and Bob Jacobsen.
> >>>>
> >>>> Building on what I have learned from this group, my mental model
> >>>> has
> >>>> arrived at the point where I believe that the mere plugging in of
> >>>> a port
> >>>> expander (like the DUAL PORT USB to Serial RS-232 DB-9 Adapter
> >>>> Cable)
> >>>> will automagically change the list of ports available to
> >>>> CommPortIdentifier.getPortIdentifiers().
> >>>>
> >>>> Experiments say this is not true however. Using same SerialDemo as
> >>>> described below, the list does not change, even after plugging a
> >>>> live device
> >>>> into one of the resulting RS-232 ports, and restarting the app.
> >>>>
> >>>> What's broken now? my hardware? or still my head?
> >>>> sj
> >>>>
> >>>> On 2006 Sep 7, at 8:40 AM, Dr. Douglas Lyon wrote:
> >>
> >> _______________________________________________
> >> Rxtx mailing list
> >> Rxtx at qbang.org
> >> http://mailman.qbang.org/mailman/listinfo/rxtx
> >>
> > _______________________________________________
> > Rxtx mailing list
> > Rxtx at qbang.org
> > http://mailman.qbang.org/mailman/listinfo/rxtx
>
> _______________________________________________
> Rxtx mailing list
> Rxtx at qbang.org
> http://mailman.qbang.org/mailman/listinfo/rxtx
>