[
Lists Home |
Date Index |
Thread Index
]
On Wednesday 30 October 2002 4:38 pm, Paul Prescod wrote:
> I've spent too long on this so I'm only going to address a few small
> issues.
>
> Alaric B. Snell wrote:
> > On Wednesday 30 October 2002 01:17, Paul Prescod wrote:
> >
> >...
> >
> > That's a different issue entirely. That's an argument for having standard
> > interfaces to data rather than anything to do with power. He's confusing
> > programming languages with data description languages. No site sends you
> > weather information in the form of a Java applet!
>
> When people (like TBlanchard) say they want to send "smart objects"
> rather than "dumb data" this is precisely what they are proposing. Not
> "applets" but Java or Smalltalk objects. And it is because they don't
> understand the POLP.
Sending objects would be fine; rather than having code that uses an XPath
like "/weather/current/temperature" you'd have
weather.getCurrent().getTemperature() - yet you could also do "JComponent
weatherComponent = weather.getGUI ()" to get the UI. There's no connection
between the dumb / smart data decision and the UI based / raw data based
decision. You could provide the weather information as a dynamic GIF
containing weather maps and little wind roses and numbers, and it'll be as
unusable as the Java applet!
> > ... They'd have to compile it
> > up again each time the weather changed!
>
> No they wouldn't. An object is a merger of data and behaviour. The data
> is typically joined with the behaviour on the recipient side.
An applet's not an object, it's a class... just the behaviour. It needs an
external method to get its data. RMI or HTTP back to the server are popular,
as well as putting paremeters in the HTML wrapping the applet.
> >...
> > And what does this have to do with the power of Scheme vs. XSLT?
>
> Scheme versus _schema languages_ (including DTDs) and pointer languages
> and CSS. Not Scheme vs. XSLT.
Oh, OK. You only need a subset of scheme for schemas, then. Define that
subset and then existing Scheme systems can process it with no changes, while
you can make limited scheme systems just for doing schema checking if you
don't need the whole thing around. So without needing an extra language you
get a purely declarative non-Turing-complete schema language... how's that?
I'd still be tempted to allow the embedding of TC predicates for advanced
constraints, though. ISTR that being suggested for one of the alternative XML
schemas, come to think of it.
> Paul Prescod
ABS
--
Oh, pilot of the storm who leaves no trace, Like thoughts inside a dream
Heed the path that led me to that place, Yellow desert screen
|