jackrabbit-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "McComsey, Doug" <Doug.Mccom...@ca.com>
Subject Bug in the TCK Tests for Same-Name Siblings
Date Mon, 12 Sep 2005 16:15:01 GMT
There is a bug in the Same-Name Siblings tests that makes the first test
looked like it worked when it actually failed and makes the rest fail
with misleading messages.

The TCK tests for same-name siblings derive from these classes:
    AbstractWorkspaceSameNameSibsTest
    AbstractWorkspaceCopyBetweenTest

The tearDown method of the AbstractWorkspaceCopyBetweenTest class is
supposed to delete the contents of the testroot of the second workspace
when the tests completes. If setUp completes and the test actually
starts then it does so. If setUp fails then tearDown does not get
called.

Because my implementation does not support same-name, I expect to fail
the tests. Instead I get this behavior.

1. The first test attempts to do the setUp.
2. It fails in AbstractWorkspaceSameNameSibsTest.setUp on this
instruction:

   if (!isSameNameSibs) {
       throw new NotExecutableException("Property
'sameNameSibsTrueNodeType' does not define a nodetype where sameNameSibs
are allowed: '" + sameNameSibsTrueNodeType.getName() + "'");
   }
3. The tearDown never executes.
4. The test ends successfully with no message !!!!!!!!!!!!!
5. The next test gets an ItemExistsException because of the nodes left
by the previous test.

I have commented out these tests as a workaround but I thought you would
like to know about the problem.

Regards,
Doug McComsey
Computer Associates,
doug.mccomsey@ca.com



Mime
View raw message