XOM API for Processing XML with Java

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

This all behaves as designed. There aren't any bugs here. It's not 
perfect, but I didn't find any solution that was. The problem this 
example demonstrates is simply that the real base URI may not be 
exposed until after the xml:base attribute has been removed. Since 
xml:base attributes are treated as attributes rather than as some 
magical thing in XOM, they can change the base URI in unexpected ways.

This example fails to address the real point of base URIs, which is 
to resolve relative URIs. If there's no relative URI to resolve, then 
this is all pointless. What's really needed is a relative URI 
somewhere in the far attribute, and then thinking about what happens 
to it at various stages including the stage where the far element has 
no parent at all. That's really why I decided to keep the entity URI 
on detachment.

What I eventually decided was the least bad option was

1. Relative URIs in newly created elements should be resolved 
relative to the parent element's base URI, rather than simply beiung 
left dangling and unresolvable.

2. Such relative URIs should not change their base simply because the 
element is moved, detached, or copied.

3. However, such elements can change their base due to the addition 
of an xml:base attribute because this is precisely what adding an 
xml:base attribute is supposed to do.

It would be possible to get a slightly more logically consistent 
position by allowing the base URI of a newly created element to 
remain null when it's appended to another element that has a base 
URI; in other words requiring the base URI on new elements to always 
be explicitly set. However, that seemed to me to violate the 
principle of least surprise for the common use cases. It required an 
extra method call to get the behavior most people were going to want 
most of the time when elements were added to one document, detached 
from that document, and then added to another. (An uncommon sequence, 
I suspect.)

Anyway, that's the best I could come up with. I don't think I can do 
better without special casing the xml:base attribute, which would 
introduce problems of its own; especially with pre-XML Base specs 
like XPath and XSLT. :-(
--

  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