RE: Strange Viewer issue

["Patrick Lidstone \(Personal e-mail\)" <patrick@xxxxxxxxxxxx>]

> -----Original Message-----
> From: Kevin Hawkins [mailto:lists@xxxxxxx]
> Sent: 02 June 2004 04:07
> To: xAP_developer@xxxxxxx
> Subject: [xAP_developer] Strange Viewer issue
>
>
>
> 	I'm having a strange problem and we had something a
> little similar at UKHA2004.
>
>  - I have Viewer running in hub mode and the Homeseer beta. I
> have a script that sends a class=message.display xAP message
> when something changes. Sure enough when it does change I see
> the message appear in Viewers window but it doesn't display
> on Watcher which is on another machine :-( - However if I
> select the message and resend it from Viewer then it does
> appear in Watcher on the other machine. What is different here ???

The sending app is different - could sending from *within* Viewer
use a different mechanism to sending *via* Viewer? (Stuart?)

> 	I ran another copy of Viewer on the PC that also had
> Watcher (the remote PC) - it saw both messages OK - I have
> copied them at the end of this message - so what was
> different ?? I wondered if Watcher was sensitive to the order
> of parameters and that Viewers hub was shuffling them - so I
> used another hub and it made no difference. I tried Patricks
> hub but it picked up a 0.0.0.0 ethernet address - now that
> got me wondering - I have a Firewire network port and the
> WiFi port on this machine showing in 'networks'. But then I
> reasoned that Viewer on a remote machine saw both these
> messages but Watcher only acted on one - consistently. Strange

It would be pretty common for an application bound to 0.0.0.0
to receive messages but not be able to send them. I'd speculate
that multiple apps bound to a 0.0.0.0 interface might all
interact normally on the *same* host, but that broadcast messages
would not escape to the LAN.

> 	Now at UKHA when P sent a message from his machine to
> my Rabbit xAP controller it didn't action it - but on another
> intermediate machine running Viewer where it did see the
> message if I resent it then worked - stranger and
> stranger.... other people could send messages to the Rabbit OK.
>
> 	So last test - I copied the two messages below to
> individual .xap files and tried sending them from the machine
> where HomeSeer failed - and both worked - so I'm not totally
> confused - anyone any thoughts ???? Could it be something to
> do with the interface or broadcast address that HomeSeer is
> using or perhaps the ephemeral port its using ???

This behaviour mirrors the same kind of erratic behaviour that
Andrew Roberts was seeing the other day with xap-desktop, xap-viewer
and xap-comfort connector. In his case, the xap-comfort connector
appeared to "go deaf" to xap-desktop, although he could see
messages
from xap-desktop in the viewer. Switching from viewer to my hub
improved, but didn't completely cure, the issue - which I find very
odd, given that my hub code is now a couple of years old, and has
never played up before, and the same is also true of the comfort
connector.

It smells to me like there might be two or more independent issues
that are compounding to cause problems. I'm not pointing fingers, but
Viewer does
seem to be a common denominator in these issues.

Kevin, do you find restarting components causes different behaviour?
If this is the case, I wonder whether the defect might be related
to not interpreting the "port" parameter in the heartbeat
correctly.
Especially as we've recently switched the ocx to using ephemeral ports.

An early problem I had with my hub was with premature time out of
heartbeats cancelling active connections. I think this would be pretty
obvious in the viewer because an item turns grey - unless it is
happening
for a really short time, between heartbeats, resulting in intermittent
dropping of messages.

Finally (not relevant to your situation) - how does xAP desktop handle
hosts with multiple interfaces? James, do you know if this code is
"right"?
Is there a way of manually verifying the broadcast address being used
(perhaps adding it to the about box or something would be a useful
diagnostic).

Patrick