Details
-
Bug
-
Status: Resolved
-
Major
-
Resolution: Fixed
-
1.2 Final
-
None
Description
Relatively shortly after introducing NamespaceResolver, Dmitri had changed it to wipe out the reverse mapping cache every time a new mapping was added. The tests, however, relied upon the last added mapping to a given namespace being the preferred prefix. The description for his change was that he was fixing a "bug", but since he didn't elaborate on what kind of bug I can't see how maintaining the cache is any more harmful than not maintaining it, and reverting this change clears up the test failures that for some reason are only seen on JDK 1.3.