Cart
[Home] [By Thread] [By Date] [Recent Entries]
Tom Bradford wrote: > > I truly appreciate your snide criticism. :-)) i honestly wasn't intending to be snide, rather to point out that XPath has been a done deal for some time now, and that it is what it is. Along the lines of "be strict in what you provide, and liberal in what you accept" I think we should strongly promote best practices for document design, yet expect that parsers dbs etc., will liberally accept data that conforms to the spec. > > The reason it wasn't there is because in order to support namespaces, > most XPath implementations require a context node for prefix > resolution. The context node is usually an element in the document that > is being queried, and it's usually an element that has been sufficiently > drilled down to so that the prefix you use in the XPath actually matches > the namespace you're looking for. Right, this is a real issue. A reasonable solution is what XPointer has come up with namely: "xmlns(prefix=http://example.com/namespace) xpointer(//prefix:example)" not that it is pretty, but does the job. i'd prefer if xpointer had provided for an xpath(//this:is/an:xpath) syntax, but that's another story. Another solution would be for the API to accept a namespace context map ... -Jonathan
|
