Re: Device discovery for control/sensor schemas

["g8kmh" <lehane@xxxxxxxxxxxxxxxxxxx>]

> Hi all,
>
> Many of the recent xPL apps are making use of the sensor/control
schemas, as these make it much easier for devices to interoperate.
>
> However, there is no way of finding out a list of valid devices
that a particular xPL connector supports.
>
> If we take DMX as an example, we can send a message like:
>
> control.basic
> {
> device=3
> type=dmx
> current=255
> }
>
> This will set channel 3 to level 255.
> But how do we know what channel 3 actually is?
> Or even if channel 3 exists at all?

..and of course even if it does exist, has it set it to 255?

>
> We should be able to be given a list of all known DMX channels, and
a friendly description of what is connected to them.
>
> The same issue applies to things like the C-Bus connector - you
shouldn't have to remember the C-Bus group address of the light(s)
you want to control - you should be able to be presented with a list
of all detected devices.
>
> The question is: Should each xPL connector be responsible for
maintaining a list of "friendly" device names, and mapping them
to
actual device names, so you'd be able to send something like:
>
> control.basic
> {
> device=lounge_lights
> type=dmx
> current=255
> }
>
> Or should it be up to a controller like xPLHal to store this
information and do the mapping between friendly devices and actual
devices?
>
With intelligent devices there may be no connector. Looking forwards
there may be semi-autonomous devices (for example the NETIOM module)
which interoperate entirely separately from xPLHal, bar
configuration. xPLHal because right now there isn't another
configurator as far as I'm aware.

My view is it is up to the device or connector to maintain the
friendly names and there needs to be some means of configuring a
device with these alias/es at startup. Therefore it would need to
announce the unit(s) it controls. DMX presents some more interesting
problems since a connector may support 512 channels (in practice
probably no more than 64) but it has no idea which devices are where.
Worse, you could add devices on channels ad-hoc and the controller
never knows.

This does present some additional onus on small devices - not perhaps
ideal but there is still no reason why a very dumb device can't
report 'switchnnn' and not be configurable and for it to be useable
in an HA system.

>
> I have a couple of ideas on how we can progress this, but I'd be
interested to hear other peoples suggestions on this one first.
>
I think universiality is the key and probably worry less about the
small devices - PIC's, etc, are becoming available with more memory -
after that it's only code!

Lehane