[
Lists Home |
Date Index |
Thread Index
]
Yah Jeni,
I am really talking about annotations at the domain level [If you were at my
presentation at eXtreme, this is all relevant to that topic]. We've got the
content marked up with annotations, and now we want somewhat better
infrastructure so as to make it easier to generate/process the information
we are producing.
I've also got the problem that the collection of annotations can grow
arbitrarily large, and many will be in separate documents.
For practical reasons, I cannot force the use of a single source document
for annotations, although I can restrict annotations to be within the
collection or parent collections.
Back to XPath-NG: The notion of XPath extensions that support nodes above
the document [i.e., collection nodes and repository nodes] would be pretty
useful. For example, we want to be able to create an XPath expression that
searches a specified collection or document within a collection.
So (in the following, treat fsname() as a name test where the argument is a
node name in a filesystem instead of a QName [so you can have spaces, et
cetera in your names]):
Collection::fsname("bar")//foo ; find all foo elements in documents that
belong to collections with the name "bar".
Collection::fsname("bar")/Collection::fsname("baz")//foo ; find all foo
elements in documents that belong to collections named "baz" contained
within collections named "bar"
Collection::fsname("bar")/Document::fsname("quuz")//foo ; find all foo
elements in documetns named "quuz" in collections named "bar".
Interestingly enough, if your filesystem hierarchy is represented by an XML
document of the form:
<Collection fsname="bar">
<Document fsname="quuz"/>
<Collection fsname="baz"/>
</Collection>
Then as Gavin would say: It's just a simple matter of transformation and
you are done. Implementation should be a cinch.
keith
|
