This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
level3/core/getbaseuri03: Even the description text for this test seems to state it is wrong. And I agree. ;-)
Request for clarification to WG (member-only): http://lists.w3.org/Archives/Member/w3c-dom-ig/2004Jan/0007.html The test and Xerces-J appears to be following Infoset which does not define a base uri property for doctype, attribute, and comment.
Created attachment 175 [details] Fixes and additional baseURI tests
Appendix C of the Core spec defines the mapping of Infoset to DOM and specifies that the baseURI of Attr is null. So nodegetbaseuri03's expectation was correct, the justification for the test in the comment was poor. I've reviewed the other tests and have added some more that reflect my first read on the Appendix. nodegetbaseuri06 tests the baseURI of a newly created disconnected element. The original and current expectation is that baseURI is the documentURI of the document used in the createElement call. The element might not be considered "contained" or "encapsulated" by the document at that point, so the formulation in RFC 2396 might not hold. It seems like a reasonable but not the only reasonable interpretation. A null baseURI for any disconnected node would seem reasonable too. I would appreciate any clarification.
Created attachment 176 [details] Slightly better this time