[xml] using schemas/xslt/xinclude to provide normaliz

I have been thinking about different options to provide xml
normalization meaning I want to be able to maintain xml files or
unique nodes with information that is reused within a specific xml
file. I gave it a try using xinclude but the parsers I have seen do
not implement xpointer() scheme but only the element() scheme. On the
other hand they do not provide way of selecting fragments using xpath
expressions as XInclude propose thru #xpointer... included in the href
attribute. Using element() I am able to maintain individual files and
include just the particular nodes I need in my final xml files,
however I have to keep the order of them because element works based
on the position of children like element(/1/2) to say use the second
child from root.

I was thinking about to use in the meanwhile (seems like xerces will
support the xpointer scheme and even an xpath scheme someday as well
as jaxen as I read from some dom4j postings) xslt to given the
specific nodes with ids just build the resulting xml file where the
nodes are repeated and I can make my xpaths simpler.

In both cases above I heve always the worry about the maintanance of
those several files that host unique nodes and I am wondering if there
is a better solution.

Still I think a better way has to exist and maybe this is using
schemas. I hope some of you could give me directions about where to
look for a good solution.

Thanks in advance
Nestor