Re: Re: Wildcarding query

[Johan Helsingius <yahoo@xxxxxxxx>]

Kevin Hawkins wrote:

> X10 is a headache in that as David says it is ubiquitous and therefore
> an X10 schema appeared for legacy support, particularly for X10
> controller applications.

Yes, it is ubiquitous, and has to be supported. But perhaps by hiding
most of the ugliness in the gateway.

> It really shouldnt have its own schema - as
> indeed most 'specific' devices shouldn't .  Aside from the device
naming
> of A1 A2 it is really problematic in that it (mostly) doesn't have
> status, or even end point presence detection, the nature of loads cant
> be determined eg dimmer/appliance

All this can (albeit awkwardly) be handled by the gateway.

> and it occasionally misses commands,
> plus all dims are relative, this makes state tracking very awkward.

Urghh, yes. But again, a "smart" gateway can compensate to some
degree.
I see the role of a gateway as a translator not only of low-level
message formats, but also between different architectures - mapping one
universe to another.

> Add the several different 'enhanced X10' specs and the different ways
people
> have implemented it eg ON at 100% vs preset dim level and it really
> becomes awkward to model.

Yes, agree. And to deal with that, there has to be a mapping
database/file somewhere , telling the software things like "H4 is a
dimmable lamp, capable of preset dim levels and smart extended
dimming".
But should this info be in the gateway, or in some central place that
also knows H4 is the light at the front stairs?


> If you name your devices appropriately in xAP and action things based
on
> functionality (eg use a  lighting schema) then it really doesnt matter
> where on a network your xAP applications sit - they will continue to
> work without reconfiguration, even if you move them.

What would be helpful is some more guidelines in using the schemas in a
smart way.

> There is no need
> to configure gateway addresses in xAP and there is no network topology
> to get in the way as xAP is broadcast in nature.

Well, I think the history of multicast on the internet has proven that
while broadcast is great in a local setting, anything bigger/more
complex will need smart routing - either manually configured, or
automatic/learned.

> Having a way to
> remotely configure applications  from one central location is
obviously
> a desireable thing....      Expect a proposal on this very soon...
> (along with a formalised groups and scenes)

Excellent!

> Johan,   tell us a little more about what you have done with xAP  -
> always interested...

Well, I haven't yet done that much. A first attempt at a "smart"
(as in
compensating-for-braindamage) X10 interface, and just some general home
automation stuff. Looking at porting/rewriting some of the Rio Receiver
stuff (as well as telephony and other sound stuff) for Linux/Unix and
perhaps Java. But I'm also involved (as board member and investor) in a
company making embedded boxes for interfacing web/internet stuff to
industrial systems and devices, and I'm pushing them to take a look at xAP.

Julf