This is the mail archive of the xsl-list@mulberrytech.com mailing list .


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

With or without PSVI


I've been trying to understand XSLT 2 (with some success). But I'm 
having trouble seeing exactly what difference the PSVI information makes.

An example. Suppose the instance document accidentally looks like this:

<foobar xmlns:xsi="http://www.w3.org/2000/11/XMLSchema-instance";
	xsi:noNamespaceSchemaLocation="foobar.xsd">
	. . .
</foobar>

and the XSLT Processor decides to accept unvalidated input, and 
exclusively uses schemaLocations to assign schema to an instance 
document (and the "type exception policy" is "flexible").

What are the conditions on the stylesheet and the instance document for 
this transformation to produce the same result as if the correct 
namespace for "xsi" had been used?

It seems to me that if you stay away from "instance of", "cast" (and 
perhaps a few other related functions/operators), and if your source 
document doesn't contain any XSD list types, you would get essentially 
the same output (thanks to "automatic" casts). You could also maybe lose 
some precision in numerical calculations if a PSVI is produced (since 
the calculations could use e.g. xs:float precision instead of xs:double).

Otherwise, all you'd get is a few warnings from the processor about 
implicit casts (if you use, say, date functions without explicit date 
constructors). But the stylesheet would retain the same functionality.

A couple of questions:

  o Am I totally wrong about this?

  o In case I'm not: Shouldn't it be possible to refactor the spec? 
Something like "If a PSVI is provided you may also use instance of, 
xf:data, and count on automagic sequence construction from XML Schema 
list types. See http://www.w3.org/TR/horrible-psvi-mess-in-xpath for 
details."

  o In case I'm wrong: Isn't it incredibly dangerous if the meaning of a 
stylesheet changes substantially in case PSVI information is provided?

  o In any case: Wouldn't it be nice to see some kind of specification 
of exactly what is required to produce PSVI-independent stylesheets? 
(and wouldn't it be nice to have an "xf:tokenize" function to be able to 
handle list types in the absence of PSVI?).

/dan
--
Dan Holmsand
dan@eyebee.com


 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]