Hi,
No, it would depend on what you added afterwards. You can add any number of elements and attributes in any namespaces. The "logical state" is identical. The namespace attrs are a description of what namespace nodes will exist in the document after parse. Namespace nodes are described in the infoset (namespace nodes are not explicitly defined in DOM, but this is illustrative). If the tree is in memory, you don't even need a prefix associated with a node in order for it to have a namespace. However, when you serialize, namespace attrs and prefixs are added so that the "logical state" can remain identical. When you give this document back to the parser, it uses the namespace attrs to associate namespaces with prefixs and nodes. At this point, the only value of the namespace attrs is to assist the serializer with serialization (for example, to force it to use a certain prefix namepsace binding as opposed to generating one as per the algorithm laid out in the DOM specs). After parse, you can move the namespace attrs round to your hearts content and not effect the "logical state".
I didn't mention the parsing behaviour, but the document creation behaviour (and only argued with the above statement). My preferred solution would perhaps be to add the xmlns attribute (is this the only one?)
If you were to try and maintain what has been described as "namespace wellformedness" it would add significant overhead for everyone and add nothing the the logical model.
Hope this helps,
Gareth
--
Gareth Reakes, Managing Director Parthenon Computing
+44-1865-811184 http://www.parthcomp.com
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
