xml-dev - Client recognition of XML Namespace prefixes

[Will Bradley <willbr@microsoft.com>]

I have a question about dealing with namespaces.  

Let's take a look at an example, and then I'll pose my question.

<root>
  <tree xmlns:dog="http://dogs.com/some/schema"
        xmlns:people="http://people.com/some/schema">
    <dog:name>Fido</dog:name>
    <dog:age>2</dog:age>
    <people:name>Will</people:name>    
    <people:age>Old</people:age>
  </tree>
</root>

When a client application recieves this xml block, and wants to interpret
it, can it, and should it, make assumptions about the namespace prefix?
Further, when calling the DOM's function selectSingleNode(), or more
specifically, when using XSL Patterns, or XPath expressions, how can one
determine the validity of a query for a specific tag?  

How do you find the tag you want within a hierarchy, without knowing what
prefix its namespace happens to be using.  It seems that there is no way to
be absolutely sure that if I search for "/root//dog:name", another app would
not have gone and switched the namespace prefix to "Dog", in which case my
search would fail due to case sensitivity, but the XML Validator would NOT
fail, because the grammar is still valid.

I realize that you can search for "/root//*:name", but that defeats the
purpose of validation, this search might return the node
"/root/tree/people:name".

Thanks in advance!

Will Bradley

p.s.  I realize I might be asking for something that is simply not possible.
The reason it might not be possible is that I (the client app), regardless
of namespace prefix, have to be certain that the Schema used to validate the
XML was in fact the correct Schema.  This would be (although not
impossible,) ... hard.