[
Lists Home |
Date Index |
Thread Index
]
On Thursday 16 January 2003 19:28, W. E. Perry wrote:
> As for 'direct
> access', it means (as it always has since I started reading code in
> 1966): conforming exactly to brittle expectations from a priori
> agreements, so that I don't have to parse it, lex it, analyze it,
> compare it to its history, or otherwise do in computerized processing
> what the human brain has to do in handling every bit of new input it
> ever encounters. Have any of the defenders of direct access here (are
> there any?) said anything to contradict that characterization?
I'll bite :-)
I say that's a purely relative statement. Whereas in a programming language
we have to access our XML via various ungainly constructs, in XSLT it's a lot
simpler because XSLT's data model is the XML data model so you don't need to
do any conversions.
However, if you have data stored in the format of a serialised Java object it
would be nice and easy to access from Java, but from XSLT you would need to
do something with an HTTP-accessible converter along the lines of:
document('http://proxy.com/JavaToXML?url=http://data.com/my-java.ser&field=foo')
The expectations some XSLT has of an XML document is just as brittle as the
expectation some Java code might have of a Java object. My "object.field"
will break if the field is renamed in the same way that "/root/field" would.
It's not about whether parsing happens or not - calling stream.readObject ()
in Java invokes a kind of parser to deserialise the object; the difference is
in data model conversion. XML is a data model that the language XSLT is based
around, while Java is based around a different one, so mapping from one to
the other involves a bit more work.
>
> Walter Perry
>
ABS
--
A city is like a large, complex, rabbit
- ARP
|
