Re: Mini-ITX PC's a the future of HA

[Marc F Hult <MFHult@xxxxxxxxxxxxxxxxxxxxx>]

On Sat, 24 Jun 2006 18:41:06 GMT, "Dean Roddey" <droddey@xxxxxxxxxxxxxxxx>
wrote in message  <6Hfng.123025$dW3.11403@xxxxxxxxxxxxxxxxxxxxxxxxxx>:


>Just for the record...
>
>One big reason our product is so good is because we are software people and
>not automation people.

Agreed.  My experience has been that neither computer hardware, nor computer
operating system, nor autonomous systems (security panel, programmable
thermostats etc) nor inter-device communication if using hardwired (IOW, not
X10, not ... add your own list here) have been an obstacle to dependable HA
for almost a decade. Most of the HW and SW concerns discussed in this thread
have been addressed solved long ago.

As Dean implies, some HA PC software for federated systems (Homeseer,
Premise systems) have/are still not reliable enough (for my requirements)
but Savoy's CyberHouse, and CQS (based on limited experience) are.

With that SW (and all others regardless of their other deficiencies) the
principal obstacle to progress is development of robust/unbreakable rules
and logic. Most of the hypothetical discussion in this thread of which
length and width of PC motherboard (or OS, or inter-device communication
method, or whatever) is "the future of HA " is jist plain silly.

Anyone who has programmed beyond (or maybe esp. including!) BASICA spaghetti
code will appreciate the difficulty of writing programs that do not contain
errors. IME, for sufficiently complex systems, writing and maintaining code
that embodies the users intentions ("rules")  is the fundamental outstanding
challenge for installer, end-user and programmer alike (even if they are one
and the same person).

In 7+ years of using a federated PC HA system, I cannot recall a PC
hardware, or OS, or (with CyberHouse) SW error that caused a significant
mistake. Other than inter-device communication errors (i.e., X-10) errors
have been predominantly of my own making by flawed/inadequate
rule-making/programming, sometimes triggered by, and compounding the
consequences of, flawed/ambiguous environmental data.

The strength of standalone components such as programmable thermostats and
security panels lies in part in that they limit the negative consequences of
bad rule sets/programming. Having written many a poor rule/line/code myself,
I have been glad that basic functionality of thermostats and security is
under the control of bullet-proof code (often ladder logic) that I couldn't
inadvertently break.

So, in my experience and opinion, the future of HA has to do with
translating human desiderata into actions and environmental conditions
through logic ("smart rules") acting on sufficiently accurate data in a way
that sufficiently reduces the likelihood of an unacceptable mistake.

IMO, most folks what git beyond the hypothesizing/hypothetical/hyperbole are
likely to arrive at a similar conclusion (for sufficiently  complex
systems).

The developers of software and human interface(s) that solve this
constellation of problems in a way that protects their intellectual property
stand a good chance at taking home all the beans.


... Marc
Marc_F_Hult
ww.ECOntrol.org