- From: Curt Arnold <carnold@houston.rr.com>
- Date: Fri, 05 Dec 2003 19:20:55 -0600
- To: www-dom-ts@w3.org
I've been running the L/S test suite against the Xerces-J CVS
implementation. Xerces-J currently fails the following tests:
DOMBuilderTest1-4: These tests execute LSParser.parseWithContext which
is not currently implemented.
HasFeature01-02: These tests fail when expecting a true value from
DOMImplementation.hasFeature("LS", "3.0") and
DOMImplementation.hasFeature("LS", null).
DOMInputSourceTest3: This test calls Document.getInputEncoding. The
Document implementation in Xerces-J has not been updated to the latest
DOM L3 Core.
DOMWriterFilterTest2: A filter is applied that should block all Text
nodes. The test expects all attribute values to have a zero-length
value, however the value is not being filtered.
The following tests were modified:
DOMBuilderTest5: Used supported-mediatypes instead of supported-media-types
DOMEntityResolverTest1: used entity-resolver instead of resource-resolver
DOMEntityResolverTest0: Expected baseURI on call to resource resolver to
be "base". There was a base (but not XML Base) attribute with the value
"base" in the source document. The appropriate value is an absolute URI
for the test6 document. Changed assertEquals to an assertURIEquals.
DOMImplementationLSTest: This test checked multiple scenarios for
DOMImplementation.createParser, some of which (asych mode) are option.
Split into DOMImplementationLSTest0-5.
I ran a coverage analysis and reported the results as bug 401
(http://www.w3.org/Bugs/Public/show_bug.cgi?id=401) and will be working
to fill in the gaps.
Oracle's 10g Java parser apparently implements an earlier version of
the L&S spec. I will attempt to hack copy of the generated test suite
to compensate so that I have two implementations to compare.
Received on Friday, 5 December 2003 20:28:48 UTC