Re: xAP/xPL and Rabbits....

["mark_harrison_uk2" <mph@xxxxxxx>]

Ian,

I'm not going to comment on XPL. However, I must comment on some of
the things you've written about xAP, since your understanding of xAP
is clearly different from mine.

Having been through some of this on IRC with others over the last few
days, it's clear that there is some misunderstanding about xAP going
around, and I want to take this opportunity to clear some of it up.

I've snipped a lot of your post, not because I don't think it's
worthy of inclusion, but to save space, and because I want to
concentrate on a limited number of points.

> xAP is intended as a broadcast only protocol, with obstacles
> against using simple messages sent from one device to another
> for control purposes.

The bit about "obstacles" is not true. It is perfectly acceptable
in
xAP to send a message which is entirely intended as a simple command
to a simple device. For instance, a touch screen could send a message
to Comfort saying "arm system".

HOWEVER, it is not acceptable for the xAP network to ONLY send that
message to the Comfort controller. There are good reasons for this...

For example, I have a Rocket controller board that gives xAP control
of me hi-fi. This board is RABBIT-based, and sophisticated enough
that it is able to keep track of the state of the alarm system. When
the alarm is armed, it never attempts to do anything to the hifi.
However, when the alarm is NOT alarmed (ie - when there's someone in
the house), it will mute the hi-fi when the phone rings.

This is only possible when all messages are broadcast. The advantage
here is that neither the Comfort alarm system, nor the touch screen
that sends the "arm alarm" message needs to be aware of the hi-fi
controller. This makes adding a new device to the xAP network very
straightforward - only the new device needs configuring.

> the underlying assumption is that end nodes have the
> intelligence to respond to the status information contained
> within messages, and act accordingly.

Again, not true. A relay controller need only have the minimal
intellgence to respond to specific commands. It is up to the end-
user / installer to determine whether they want a simple relay
controller, and move the intelligence outboard to some more
sophisticated controller, or whether they would rather have a more
sophisticated relay controller capable of handling the IF/THENs
itself and respond directly to "status" messagess.


> xAP's initial design goal calls for a fully distributed control,
> without a central controller. This goal has softened as the
> harsh realities of implementing real world systems have arisen.

The use of a PC in _my_ installation serves two purposes:

- Providing a web controller

I don't need one, but I want one :-)

- System configuration

The second use is akin to using a PC with a Pronto - you use the PC
to _set_up_ the devices, but can then unplug the PC and take it away.
There is no central controller required to do things like mute the PC
when the phone rings.

However, there is no OBJECTION to using a central controller with
xAP. It's just a question of what the end-user / installer is more
comfortable with. Given that most HA systems are coming at this from
a controller-centric mentality, xAP allows a comfortable migration
path retaining the use of the central controller software on day one
of installation, and migrating the intelligence out to the nodes as
the user becomes more familiar whole mindset of distributed
intelligence.

> Heirarchical Message Schemas
> xPL mandates certain basic behaviours from each message class.
> Whereas the relationship between xAP classes and types remains
> open to developer interpretation

Yes and no. The xAP group maintains a list of "official" schemas,
and
welcomes submissions from developers. That means that anything like,
to pick three at random:

- relay control
- X10
- Audio control

... is implemented in the xAP world in a consistent manner.

As such, it's EASY to, say, take out a SliMP3, and put in an
Exstreamer - or vice versa, depending on what we all discover at the
UKHA2003 listening tests :-)

Regards,

Mark Harrison
xAP Project Manager