These are the test cases failing on bridge. There were 15 failures, but I
think are only 5 issues (some issues give rise to more than one failure). We
should get rid of them, somehow. Please comment.
run.bridge:
[java] There were 15 failures:
[java] 1)
testGetValue(org.mmbase.bridge.EmptyNodeTestTransaction)junit.framework.AssertionFailedError:
Empty byte field did return null (should be empty byte[])
An unset byte-array field returns 'null' now and not an empty byte[]. This
fails every time it is tested.
Should we change the code or the test-cases?
[java] 8)
testCreateAlias(org.mmbase.bridge.EmptyNodeTestTransaction)junit.framework.AssertionFailedError
[java] at
org.mmbase.bridge.NodeTest.testCreateAlias(NodeTest.java:177)
This seems a bug. I propose to fix it. Somehow the alias is not found back
immediately after creation.
[java] 9)
testSetContext(org.mmbase.bridge.EmptyNodeTestTransaction)junit.framework.AssertionFailedError:
java.lang.RuntimeException: node number was invalid(-107)
Connot set context on uncommited node, or something like that? I propose to fix
that.
[java] 12)
testGetListWithInvalidParameterStartNodes(org.mmbase.bridge.CloudTest)junit.framework.AssertionFailedError:
Should raise a BridgeException
The following things are succeeding now:
nodeList = cloud.getList("" + aaNode1.getNumber(), "aa,bb", "aa.bytefield", "", "",
"", "", false);
nodeList = cloud.getList(null, "aa,bb", null, "", "", "", "", false);
nodeList = cloud.getList(null, "aa,bb", "", "", "", "", "", false);
The test-case is not specifying why it should fail. Perhaps empty strings
should not be accepted but are now. In which case I suggest we change the
test-case, because that seems not so very bad, and useful in some cases.
Unless someone has an argument _why_ empty string may not mean the same as
null.
[java] 15)
testCreateRelations(org.mmbase.bridge.RelationTest)junit.framework.AssertionFailedError:
related list has size 0 but should be 1
This comes from the following code:
//we now have 3 node
//aaFirst -> related -> bb -> related -> aaSecond
//check back if we can get the direction
NodeList aaRelatedList = aaFirstNode.getRelatedNodes("bb","related","source");
assertTrue("related list has size " + aaRelatedList.size() + " but should be 1",
aaRelatedList.size() == 1);
I think this is a test-case error, since the 'bb' node is on the
destination side of the aaFirst node, so getRelatedNodes("bb", "related",
"source") should indeed give back 0 results, and not 1, as the test-case was
asserting.
I propose to change the test-case, but I don't understand how this could
ever have been succeeding.
Please, before the 1.7 release the above issues should be settled.
Michiel
--
Michiel Meeuwissen
Mediapark C101 Hilversum
+31 (0)35 6772979
nl_NL eo_XX en_US
mihxil'
[] ()
