XOM API for Processing XML with Java

[Elliotte Rusty Harold <elharo AT metalab.unc.edu>]

I'm continuing to think about this. It occurs to me it might be 
helpful to consider the use cases. In particular, if someone does add 
a new element with no explicit base URI to a previously parsed 
document what are they likely to want?

This only really matters at all if that element somehow contains a 
relative URI, e.g. in an href attribute. Otherwise the question is 
moot.

If it does contain a relative URI, then I think they expect that URI 
to be resolved relative to something and the only plausible thing for 
it to be resolved relative to (assuming no explicit base URI has been 
set, in which case, again, the question is moot) is the base URI 
inherited from the parent element and its ancestors.

Therefore, it makes sense that when the element is detached it 
retains the base URI it would have had in the original document, even 
if it was created from memory.

On the other hand, if the element is then added to another document 
with a different base URI, and then a child with no base URI is added 
to the detached element, it may be a surprise that relative URIs in 
the child are resolved relative to the old document and not the new 
one. Damn. Thought I had the answer for a minute. Damn. This is a 
real 51-49 situation. I think the most common use case still says all 
nodes should retain their maximally absolutized, actual base URIs 
when detached. But it's hardly the only use case. I just wish I could 
find one really convincing argument that demonstrated clearly what 
the right answer was.
--

  Elliotte Rusty Harold
  elharo AT metalab.unc.edu
  Effective XML (Addison-Wesley, 2003)
  http://www.cafeconleche.org/books/effectivexml
  http://www.amazon.com/exec/obidos/ISBN%3D0321150406/ref%3Dnosim/cafeaulaitA