Well, to quote from the protocol document:

	4.      X-10 Reception.

	Whenever the interface begins to receive data from the power-line,
it
	will immediately assert the serial ring (RI) signal to initiate the
	wake-up procedure for the PC. Once the data reception is complete,
the
	interface will begin to poll the PC to upload its data buffer
(maximum
	10 bytes). If the PC does not respond, then the interface's data
buffer
	will overrun, and additional data will not be stored within the
buffer. 

So, in this case, it shouldn't make a difference since we are using a
dedicated process on a system which never sleeps. (I assume RI was chosen
for compatability with "Green" PC motherboards.)

The simplest solution appears to be the one suggested by Allen Briggs et.
al.: Set "local" in /etc/ttys and don't open the port with O_NONBLOCK. It
appears to work, though I'm still trying to run down one protocol issue
(topic for comp.home.automation).

Thanks...
<<<John>>>

"One World, One Web, One Program" - Microsoft Promotional Ad
"Ein Volk, Ein Reich, Ein Fuhrer" - Adolf Hitler

> ----------
> From: 	Erik E. Fair[SMTP:fair@clock.org]
> Sent: 	Friday, August 07, 1998 11:32 AM
> To: 	Ruschmeyer, John
> Cc: 	port-sparc@NetBSD.org; tech-kern@NetBSD.org
> Subject: 	Re: Blocked I/O on serial ports
> 
> See
> 	ttyflags(8)
> 	ttys(5)
> 
> In /etc/ttys, you can set the state of a port (e.g. "local"), and
> ttyflags(8) gets invoked at boot time. So, this is partly a system
> configuration issue; you can change it and run ttyflags yourself (so that
> you don't have to reboot).
> 
> The other question is whether you've got the serial cable wired right; I
> don't think that the Sun serial ports are wired to do anything with RI -
> what does the X-10 box use it for?
> 
> 	Erik <fair@clock.org>
> 
>