RE: Schema definitions -- a discussion

["Ian Lowe" <ianlowe@xxxxxxxxxxxxxxxxx>]

Can I just say...

Absolutely.

I think that this is one of those Big Ideas(tm) that change things pretty
radically. It seems to me that this is just so elegant, powerful yet
inherently simple that it just obviously has to happen. :D

I have seen it myself plenty times as well - I built a simple switchplate
app, and hit the same problem you describe - the developer has to code in
the schemae to be used.. And that limits the re-use of the app.

> First, let me be clear in that I'm not suggesting altering the
existing
elements of the plugin files.

I wouldn't try to limit the usage by shying away from changes to the plugin
architecture - it seems pretty logical that a schema definition for (for
example, X10) contains a lot of the information that a developer would have
to add to their own fragment for an X10 application.

It might give us a way to simplify things - provide a "lowest common
denominator" that an application would have to honour, then the
developer
can add specialisations or extra features into their own fragment that are
unique to their own version.

> Like plugins, there would have to be some "master index"
that could be
consulted to find where to
> fetch schema documents automatically.

Which, in a sense would give us an xml representation of the whole
protocol!

>I also think the idea of a limited form of "inheritability"
is needed

Agreed - it also gives a nice mechanism for developers to test out new
schema

>I would be happy to supply a tool that allowed editing these documents,
much like XPE allows editing plugins.

Let's see what the others say - I'm sold.

> First question, I suppose, is -- is there any interest in this?

That's a yes. Oh yes.

> Next: whats missing?  what is too convoluted or not described enough?
etc.

I'll bow to greater knowledge on that front - I'm not that great at reading
XML at the best of times!

Ian.