"Craig Buchanan" <msnews.microsoft.com> wrote:

> assume that i have this xml stored in a domDocument:
> <Persons>
>     <Person>
>         <FirstName>Craig</FirstName>
>         <LastName>Buchanan</LastName>
>     </Person>
>     <Person>
>         <FirstName>Another</FirstName>
>         <LastName>Buchanan</LastName>
>     </Person>
> </Persons>
> 
> what is the best mechansim to be able to "query" this document using
> selectsinglenode, and always return a <Persons> element and the 0 to many
> <Person> elements that match the xpath query?
> 
> Case A: if I use "//Persons", i get the xml listed above.
> 
> Case B: if i use "//Persons/Person[LastName='Buchanan']", i get:
>     <Person>
>         <FirstName>Craig</FirstName>
>         <LastName>Buchanan</LastName>
>     </Person>
> 
> it should return the entire xml string listed above.
> 
> Case C: if i do "//Persons/Person[FirstName='Craig']", i get:
> 
>     <Person>
>         <FirstName>Craig</FirstName>
>         <LastName>Buchanan</LastName>
>     </Person>
> 
> i should get
> <Persons>
>     <Person>
>         <FirstName>Craig</FirstName>
>         <LastName>Buchanan</LastName>
>     </Person>
> </Persons>
> 
> what am i missing?

I'd assume that selectsinglenode does exactly what its name implies,
and adds an implicit [1] after every XPath selection, and for any
non-root match, it returns the containing element plus its parent.

What does it do if <Person> is 45 element levels deep in a big 
hierarchy, with assorted ancestors which may have compulsory elements 
or mixed content? Should it just return a well-formed document with
minimal element markup all the way back to the root element every time?

///Peter