OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [xml-dev] PSVI using existing infoset items

Elliotte Rusty Harold wrote:

> At 7:33 PM +0200 10/16/01, Eric van der Vlist wrote:
>>Let's take a very simple example...
>> <bar>...</bar>
>>What I had in mind is something that could be processed exactly like the original document with some information added, something such as (very rough draft):
>><foo xmlns:...>
>> <psvi:element type="complex" datatype="foo" whitespace="false"/>
>> <bar>
>>   <psvi:element type="simple" datatype="xs:string"
>>       primary="xs:string"
>>       minOccurs="0" maxOccurs="unbounded" whitespace="false"/>
>> ...</bar>
> This is still too complicated. In particular, it changes the parentage of elements and tree structure of an XML document. An XSLT style sheet that operated on the original document would likely fail when applied to the PSVI document. What's needed here is a pure attribute solution so that PSVI information can be merged with the original document without changing its basic structure. For example,
> <foo xmlns:psvi="..." xmlns:prefix="..." psvi:type="prefix:typeA">
>   <bar psvi:type="prefix:typeB">
>    ...
>   </bar>
> </foo>

That's what I had in mind first, but it won't work to qualify attributes 
(I wish XML could have adopted the principle of equivalence between 
elements and attributes taken by RDF...).

To qualify attributes, we need more flexibility and this flexibility can 
only be obtained by an element!

Even with what I am proposing, most of the transformations would work 
assuming that you add an empty template for the elements from the PSVI 
namespace. Also, keeping the content model as element only or empty will 
insure the the string value won't be affected.

> Most applications that were designed to process the original document could also process the PSVI document without being rewritten.
> I also don't think you need anything more than the qualified name for the type of each element. If an application wants details about the type, then it should look at the schema. 

Not if the purpose is to represent the full PSVI to avoid having to 
modify the APIs to use it.


Rendez-vous  Paris pour le Forum XML.
Eric van der Vlist       http://xmlfr.org            http://dyomedea.com
http://xsltunit.org      http://4xt.org           http://examplotron.org