On 7/16/06, Jirka Kosek <> wrote:
> My own understanding is that attribute is always related to an element
> on which it appears. So you can know identity of unnamespaced attribute
> if you take current element into account, and you will save few bits of
> memory when parsing and processing XML.
If it was like this then it was fine. But I think the spec doesn't implies this.
Please consider this XML:
<a xmlns="http://my.com/namespace">
<x xyz="hello" />
</a>
for element a, the namespace declaration is present for the null
prefix. This is known as the default namespace.
Now what namespace binding applies to element x and attribute xyz. To
element x, the default namespace applies (http://my.com/namespace).
But attribute xyz is in the null namespace (i.e. there is no namespace
binding applicable to it). This is what that looks illogical to me.
Regards,
Mukul
