Hello everyone!

I was hoping someone could clarify to me the difference between these two, in TDS schemas (not XSD).
Basically I am defining an element with an attribute.


In GUI, this was generated by inserting a complex element to the schema and then adding an attribute:

<xs:element name = “full”>
xs:complexType
<xs:attribute name = “width” type = “xs:positiveInteger” use = “required”></xs:attribute>
</xs:complexType>
</xs:element>

----------------------

In GUI, this was generated by inserting an “element with attributes” to the schema and then adding an attribute

<xs:element name = “full”>
xs:complexType
xs:simpleContent
xs:extension
<xs:attribute name = “width” type = “xs:positiveInteger” use = “required”></xs:attribute>
</xs:extension>
</xs:simpleContent>
</xs:complexType>
</xs:element>


What I would liek to know is why these two exist? Is there a difference in terms of speed or anything else when Tamino validates the node against an instance?

I am evaluating a software called “Stylus Studio”, so that I could print these schemas graphically (I own XmlSpy, but that one doesn’t print attributes, period!), and the software won’t graphically render attributes if they are declared within xs:extension. So I am thinking I will use only complex elements from now on and not the “element with attributes”, if, there is no real difference.


Thank you very much

Peter Endisch

---------------------------------------
I’m a Zen Garden Maintenance Engineer
---------------------------------------

Hello everyone

I poked around a lot and I guess I answered my own question. I thought I’d share it with you, in case it helps someone. Furthermore, if I am wrong about my assumption, you are more then welcome to correct me. I don’t claim to understand everything perfectly.


A complex element cannot really have a “value”.
You can force it to accept a “value” by using “mixed:true”. By doing that, you can have both subnodes and a textual value.
ex:


textual value, thanks to mixed:true
subnode-value


However, by doing this, you cannot declare a datatype for the textual value. This is, obviously a huge problem, and I think therefore that mixed:true should be used sparingly (for instance, if you want to have a node that contains a description (xhtml), and you don’t know how many subnodes and or text you will have).

On the other hand, “element with attributes” is great when the node you’re using is a leaf node and it will also have attributes. You can of course declare a datatype for the value of the node.

Lastly, if you want to have a node that has no subnodes, and has NO TEXTUAL VALUE of any sort, but you want it to have attributes (great for nodes that have attributes representing foreign keys), then I find using a complex element with attributes, and no subnodes and mixe:false is the best thing.
That way, the schema doesn’t allow any value for the node (just the attribs).

Any thoughts?

Peter Endisch

---------------------------------------
I’m a Zen Garden Maintenance Engineer
---------------------------------------