As I am building my house I find it very hard to mix and match HA devices in the US.  For example, I want to use European KNX panels (See here), why? because they look good, with hardware I can find in the US (e.g. Lutron).  Here is the kicker, the installers I talk to immediately reply "But you have to go to Crestron".  Mind you all I am trying to do is lights/shades/AC, I point out that I don't want to spend 100k in integration just for these 3 functions of integrated luxury.  The choice of wall panels is poor at best and I don't want to spend a fortune to get a non-programmable wall-panel black box solution. In fact that is OR's primary raison d'etre, to develop wall panel technology that is open, free and doesn't suck.

In Europe there is this standard called KNX and by standardizing the bus they open up the panel field to independent panel vendors, they create a market.  Very much like Java EE did in IT.  That means there are people competing on panels looks alone and the result is a lot more eye pleasing than the Crestron equivalent for example. Competition makes this market better.  Of course mind-set right now is "closed is good", they want to own the whole platform.  What a cro-magnon approach to markets.  I was thinking I could bring a KNX sensor network, map it to a US hardware layer like Lutron, through OR.  I will investigate but I suspect it won't be practical to hire European electricians to build in the US.

I am already using OR as the integration layer and panel generation layer.  The idea is that I want a native hardware layer of switches as a redundant physical layer of control.  If OR goes down, the wall switch is always available.  I am gravitating towards Lutron HomeWorks, because it has a good rep, does shades and lights, OR will get software to talk to it and it is a high-end offering that has a reputation for reliability.  Any thoughts or recommendation on hardware are welcomed, btw.

But it also became very clear to me that the well worn IT development approach to implementation could be a logical next step for the industry.  The problem I just described, that it is hard for me to get the custom *switch* panel in my future home, can be recast in Model View Controller terms.  Very simply there is no portability of View.  Most HA professionals will be readily familiar with the controller layer.  It is the Crestron/OR CPU at runtime.  What is missing is pluggable views.  Views that are easy to program, deploy and have rich information.  For that the controller layer needs to expose state in a readily consummable way.  Consider that the predominant HA protocol, IR and X10 are stateless and one way and you will understand why state of the art panels in the field advertise "stateful screen".  Of course rendering panels for control in HTML, iPhone, Android would be the "view" layer. The controller would just listen for view commands.  This is the way OR 1.0 works today btw.  We use REST interfaces on the controller that we call from the view.  We specify a XML screen format (just like android does) and each view command sends a HTTP request with the command encoded.

But for now, I suspect I will have to settle for separate control wall switches and a OR integration layer on iPhone or perhaps android in the future. My wall switches will look like wall-warts but there doesn't seem to be a choice for me, is there?