Re: HA Object Model or Taxonomy?

[alan.trevennor@xxxxxxxxxx]

Hi Ake,

Thanks for bringing this to our attention. I've had a look at the Wiki and
=
read through the Primer and found it very impressive. I think I have a
rudi=
mentary understanding of how VSCP differs from what we have been
discussing=
here. Let me see if I can test that understanding and explain to everyone =
else on this board that's interested.

In short, our discussions here have been about the possibility of making
an=
object model that can classify household fittings, appliances and faciliti=
es in a way that not only groups them together logically, but also
provides=
an infinitely extensible mechanism for adding new innovations and variatio=
ns as they appear. In such a model, each object encompasses technology
piec=
es that form a very specific functional entity or product (such as a
centra=
l heating system or a doorbell).

So what about VSCP:

Firstly, all VSCP devices live on a common network (which can be TCP/IP
ove=
r Ethernet or wireless or it can be a CAN (Controller Area Network) bus
set=
-up, which carries DC power as well as data to and from devices using a
CAT=
5 cable. Importantly, each node in the HA system "hears"
everything that al=
l the other nodes "say" and can decide whether or not to take
notice of it =
- very much like RS485 systems in manufacturing environments, or the CAN
bu=
s in a modern car, bus or truck. VSCP works over all these (and other)
phys=
ical level networks, and you can have have mixed types in one setup if you
=
need to.=20

The VSCP approach to dealing with functional equipment and fittings seems
t=
o me to be more reductionist and event driven. With VSCP You say, okay,
her=
e is Gavin's multifunction MP3 Player and Alarm clock: It has buttons
which=
can be pressed, it can respond to events to play music, it can be sent a m=
essage to tell it the current time etc.=20

But, VSCP has no idea what function this device serves, only that it
genera=
tes events (which have to conform to the class and event codes prescribed
i=
n the VSCP specification, as listed on your Wiki). VSCP doesn't care that
t=
his device is also capable of responding to events (though it can request
d=
etails of such capabilities and display them) because it's up to the
device=
s themselves to pick and choose what events will trigger whatever actions
t=
hey are capable of. Each device has a DM (Decision Matrix) and this is
stor=
ed ON the device itself, not on a PC or supervising node. The DM maps
recei=
ved events to actions. Events are serviced in an almost totally
mechanistic=
way, with no need for help from a central controlling node, once the syste=
m is set-up. DM's can be downloaded to a management program running on a
PC=
(Linux or Windows) that makes it easy to edit the DM and reinstall the mod=
ified version, but that's all, the PC is not used a "run time"
(so to speak=
!).

I think that, in terms of what it achieves, this is not too different from
=
what we had already been discussing here, except in one (perhaps)
important=
regard: We'd kind of been building a view that somebody looking at the tra=
ffic between our objects (supposing we ever built our version) would be
abl=
e to tell from the object type codes that a button press originated from a
=
very particular device type; say, an MP3 player with a built-in alarm
clock=
.=20

Further, they would be able to find out by consulting our model, that this
=
very particular device was related to clocks and was an appliance. In the
V=
SCP model there is no need for any such knowledge: It's simply that button
=
"x" has been pressed, so "relay 99" must be turned on.
End of story.

Rather than having a model that tries to embrace all possible combinations
=
or consumer goods and equipment, it seems to me that VSCP has taken the
rou=
te of breaking down things into more or less indivisible technology
chunks:=
A Button that can be pressed: An information source that outputs a tempera=
ture: A ringing indicator for a phone (but we don't know what kind of
phone=
or where that phone is located, and we don't need to): A doorbell button h=
as been pushed but we don't know what door, we only know when the event
hap=
pens we have to generate send a "ring" event to another node (or
several no=
des). And so on....=20

Now, if I have understood the fundamentals of it properly, I find the VSCP
=
approach very attractive (and I'm very sorry that I didn't find it earlier
=
when I was looking around). It does differ a little from what Gavin and I
h=
ad been discussing, but I think maybe it achieves the same goals, just in
d=
ifferent ways to those we were contemplating.=20

I think I should let this new information percolate for a while: I will
rea=
d your spec some more, and try out some ideas on pencil and paper to see
if=
they work on VSCP or whether there are things I would want to do that woul=
d be difficult. I haven't thought of anything that could not be done so
far=
, which is a good sign! Besides, I see you boys have been doing VSCP for
so=
me years, so I guess you've found ways to accommodate most requirements by
=
now!

I will come back to you with further thoughts in a week or so. In the
meant=
ime, thanks again for bringing VSCP to our attention.=20

Alan T

P.S> I was happy to see that, on the VSCP introductory page you too were
co=
mplaining about how all these standards bodies and so-called
"open" specifi=
cations are actually only for the rich and the corporate! Totally agree:
It=
's one of the main barriers to HA taking off and it should not be like
that=
.=20

--- In ukha_d@xxxxxxx, YAP <x112358@...> wrote:
>
> Hi,
>=20
> I'm the maintainer of the VSCP protocol (Very Simpe Control Protocol)
> and I claim we have such an object model. We are not finished  by far
>  but we have the model or I rather like to call it the abstraction.
> Everyone wanting to participate in this is welcome.
>=20
> What we have is rules for how
>=20
> - a button in a webpage, or on a pc, or on a physical board can switch
> something in a remote location and know it it was switched on or not.
> The actual button can actually be controlled by any protocol
> available.
>=20
> - A slider on a web page, on a pc on a physical board can dim a lamp
> in a remote location and know that it was dimmed. The button can
> actually be controlled by any protocol available.
>=20
> - A 1-wire sensor can be switched for a x sensor out at a remote
> location without any changes to the application software that draws a
> diagram from it's measurements, control something etc.
>=20
> - Every device can be configured in the same way. Every device can
> tell you how you configure it.  Every device know where it's manual
> is. Every device have a globally unique id.
>=20
> - Every device x can...  Every controlling application... In other
> words it's a lot.
>=20
> - Can be used on very low cost devices but also on high en devices.
> Very small footprint for protocol implemention.
>=20
> - You can build autonomous system like the first video  here
> http://www.vscp.org/docs.php or you
can control it with a pc, table,
> phone or whatever.
>=20
> - Firmware can be updated remotely.
>=20
> - Works on any platform. Windows/Linux and Mac with some minor work.
>=20
> - Is open, its free. No fees for usage. No restrictions on
> incorporating it into your own products.
>=20
> - You can use your old protocol centric apps like X10, xAP as before.
>=20
> etc etc
>=20
> You can find more information here (http://www.vscp.org/) and some
> products here (http://www.auto.grodansparadis.com/).
If you dare to
> take on the specification its here
> http://sourceforge.net/projects/m2m/files/VSCP%20Specification
>=20
> We do not claim to be THE protocol we just want to be THE abstraction.
> Just because we can support all other protocols if you can coup with a
> few roles.
>=20
> The work we now at last concentrate on is user interface. A lot of
> things will happen in this area in the near future. HTML5 widgets and
> some phone app. stuff. Funny thing.  We need more people taking part
> in this and who wants to be part of something that will change the
> world as we know it.
>=20
> Cheers
> /Ake
>=20
>=20
> --=20
>
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> Grodans Paradis AB - Brattbergav=E4gen 17, 82050 LOS, SWEDEN
> Phone: 46 (0)8 40011835   VAT no/orgnr: 556855-47777
> http://www.grodansparadis.com  http://shop.grodansparadis.com
> http://vscp.org  http://shop.grodansparadis.com
> Musique: http://www.soundcloud.com/hedman
>
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
>




------------------------------------

<*> Join the Automated Home Forums
http://www.automatedhome.co.uk/vbulletin/