XML DOM Q: understanding namespace nodes

[keshlam@xxxxxxxxxx]

> I was trying out Khun Yee Fung's XPathTool and am noticing it only
>lists namespace nodes where they are declared.

I can't address that tool, but I can address the DOM. Important point
before we start: Remember that the official position of the Namespace folks
is that the namespace prefix is only "syntactic sugar", and that the
Namespace URI is its real "name"; some of them have argued that prefixes
should not be visible to programs at all, and that users should not assume
that an XML document when written out will have the same prefixes, declared
at the same places, that it did when it was read in.

The DOM WG attempted to compromise between that view and the desire that
programs be able to avoid making completely gratuitous changes to the
document. Hence:

DOM Level 2 performs "early binding" of namespaces to elements and
attributes. The namespaceURI is set at the time namespaced nodes are
created and can not be changed thereafter. As far as the DOM is concerned,
the prefix field of a node is just retained as a hint for serialization; it
_can_ be changed, but changing it has no effect on the node's identity.

Note that the DOM does not currently enforce namespace well-formedness; if
you start with attributes foo:x and bar:x, and change both their prefixes
to baz, that may make trouble when you attempt to serialize the document
back to XML syntax but it's perfectly legal as far as DOM Level 2 is
concerned.

The DOM treats namespace declarations as ordinary Attributes, appearing on
a single node. They are not "inherited". (Inheritance was considered, but
was deemed inappropriate in an early-binding model. If you want to know
what was defined by your ancestors, look upward.) Parsers are expected to
recognize them as they go by and use that information to set the
namespaceURI field properly on nodes with that prefix created at or below
that point in the tree.


DOM Level 2 does not provide any built-in mechanism for looking up the
association between prefixes and their declared URIs. It's fairly likely
that Level 3's "content model" chapter, which also addresses namespace
well-formedness checking, will define such a lookup. It was deferred
largely because the early-binding behavior makes the exact definition of
this lookup somewhat less well defined, we didn't have an answer ready in
time for Level 2's release.



Reminder: DOM Level 2 is currently in Candidate Recommendation phase. You
still have a few more weeks in which to start building a prototype and tell
us whether anything is underspecified, incomplete garbled, or impossible to
implement efficiently. (Note that quibbles with the architecture should
have been submitted before Last Call ended; the question now on the table
is whether folks can really build and use what we've described.) After
that, assuming no show-stoppers arise, we go to PR and voting begins.

______________________________________
Joe Kesselman  / IBM Research



 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list