Re: [Xmlastro-discussion] Weather, etc. Was: Multiple Objects

[Tom W. <dev...@to...>]

On Wed, 2 Apr 2003 06:05:07 -0500 (EST), Peter C. Chapin 
<pc...@ec...> wrote:

> I'm not familiar with <xs:anyType>.

It is the root of the datatype hierarchy - <element type="xs:anyType"/>.

> I do have a reference that talks about
> the <xs:any> element, however. It gives an example like this:
>
> <xs:element name="XHTMLSection>
> <xs:complexType>
> <xs:sequence>
> <xs:any namespace="http://www.w3.org/1999/xhtml"
> minOccurs="0" maxOccurs="unbounded"
> processContents="skip"/>
> </xs:sequence>
> </xs:complexType>
> </xs:element>
>
> This apparently allows an <XHTMLSection> element to contain any elements
> from XHTML. The processContents attribute controls how the contents of
> that element are validated. The value of "skip" means no validation, but
> other options are possible (namely "strict" and "lax").
>
> Is this the sort of thing we're talking about?

Yes! <xs:any> and <xs:anyAttribute> allow for any content to be extended to 
the schema. The difference with <elem type="xs:anyType"> is that the 
element name is defined and it is only the content of the element itself 
that varies. It works in a similar way to <elem xsi:type=".."> in the 
instance document.

For example:
A schema doc defines weatherType, a complexType instantiated as
<temperature>
  <value units="C">20</value>
  <notes>Roll on summer</notes>
</temperature>
in the namespace http://weather.sourceforge.net

And in the astroXML schema within <observation> the temperature element is 
defined as
<xsd:element name="weather" type="xsd:anyType" minOccurs="0"/>

The instance would then be:

<observation xmlns="http://xmlastro.sourceforge.net"
		        xmlns:w="http://weather.sourceforge.net"
			  xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
			  xsi:schemaLocation="xmlastro.sourceforge.net astroxml.xsd
			   			    http://weather.sourceforge.net weather.xsd">

<weather xsi:type="w:weatherType">
	<w:temperature>
		<w:value units="C">20</w:value>
		<w:notes>Roll on summer</notes>
	</w:temperature>
</weather>
			    </observation>

As with <xs:any>, the weather schema is added at instance creation, so 
different content models for weather can be used, with the only difference 
being the element itself <weather> has been defined in the astroXML schema.

> Note that apparently the namespace attribute can be left off so that the
> content of the element can be from any namespace. The reference I have
> even says that this is a nice thing to do during development for exactly
> the reasons we have been discussing. Using this approach does sound good
> and it may be a good way to handle <notes> permanently.

<xs:any> has the advantage of allowing any content, so I agree that during 
development this is probably a better option and more extensible. Using 
<xs:any minOccurs="0" maxOccurs="unbounded" namespace="##any"/>
within the content model for observation we can integrate all the 
additional content models for person, location, weather etc, and then 
decide on naming namespaces at a later date.

But then as we progress, we are probably going to want to tie down the 
content more. The idea behind xs:anyType was to allow an author to define a 
more advanced weather definition than the one we offer by default, without 
interferring with the actual astroXML schema.

Besides, as far as development goes, I was somewhat aghast to discover that 
XMLSpy will validate simpleTypes using the xs:anyType method above but 
*not* complexTypes!! XSV does validate both types.

Tom