> > The IdResolver should be modified to include this namespace URI in the array > at > line 164, and the code at line 266 should be updated to differentiate the 2 > namespaces. > >
Thanks for looking at this. With the above, I didn't mean to imply those line references were the *only* things that needed to be changed, however. I think (unfortunately) there are some hardcoded assumptions about the length of that 'namespaces' array in the code between those two line references. It's a bit hard for me to grok what the code is trying to do, it's a bit convoluted, but: Isn't the idea essentially that after the call to getEl() in getElementBySearching() the 'Element [] els' array contains the resolved element in the slot (index) that corresponds to the same index in the namespace array for the namespace from which the ID attribute came? And the 'els' array is one slot larger than the namespaces array, so that the last slot just means "no namespace" (happens if the attribute local name is just "ID" , or "Id" or "id", with some exceptions for namespace-qualified attribs that use those). If so, then the code is now a little off - the array size references need to be incremented, or even better, changed to be relative to the length of the namespaces array (actually visible as static List 'names'). Or something along those lines. See for example: Line 176 Element []els=new Element[6]; Line 230 elementIndex=(elementIndex<0) ? 5 : elementIndex; Line 236 index=(index<0) ? 5 : index; Line 250 index=5; Line 256 index=5; The namespaces array did have length 5, now it has length 6 - so the "no namespace" slot in the 'els' array (which should now have length 7) would now be 6 , not 5. Again, it would seem to me to be better to make this all relative to the size of the namespaces array in the first place, not absolute. Those are just the things I see. I'm not completely clear on the search algorithm, so don't take my word for it. Thanks, Brent
