RE: Re: xPL announcement/description protocol -- was xPLDiag

["Ian Lowe" <ian.lowe@xxxxxxxxxxxxxxxxxx>]

Okay,
One point worth making here as we discuss this is that this conversation
has happened before - in fact, it's so eerily familiar that I was
getting dejavu - I dug back into the archives a bit, and sure enough,
the same concerns were voiced about xplhal, and the xplhal manager.

Here's the thing - John B proposed a change to the control.basic to
allow for a control.request message... The point of which being that
apps with more than one "device" behind them (like the CBUS
interface)
could report back the devices they know about...

To me, this logically fits the suggested about. Schema usage so well...
At the time, everyone agreed that it was good... But nothing was
formally changed.

I believe that the protocol took a great step towards maturity when
xplhal manager and the configuration process were hammered out - I can
see us on the verge of something similar here.

One of the things we "old skool" guys have to remember is this -
the
schema have been changed to add extra fuinctionality before, even the
hbeat.basic one. Until the hbeat.end extension was added, hbeats didn't
explicitly *end* - they just timed out.

That being said, several extensions of the protocol were discussed
before (and since) and have not been adopted - largely because the
authors came up with alternative ways to accomplish these things without
protocol changes, simply by using schema in more intelligent ways.

> It's all very well trying to stick to the simple old ways, but if it's
holding back
> making the system more user friendly, then we're just shooting
ourselves in the foot by > not biting the bullet and improving the way
these things work.

And this is the crux of the matter - personally, I have had a week of
things being a bit unstable whilst testing new stuff, so I'm yearning
for a bit of "just works"... But at the same time, we *need* to
be more
user friendly.

> Glossy front ends and simple diagnostic tools are what you need to
attract a larger less > code-headed user base.

Absolutely - I don't think there's any disagreement on that front, only
on how to get there!

Ian.