On 15.1.14 2:44 , Jukka Zitting wrote:
Hi,
On Wed, Jan 15, 2014 at 7:46 AM, Marcel Reutegger <[email protected]> wrote:
this change causes a test failure:
Failed tests:
testLockWithPendingChanges(org.apache.jackrabbit.test.api.lock.LockManagerTest):
Attempt to lock a node with transient modifications must throw
InvalidItemStateException.
Oops, sorry about that. I failed to run the build with
-PintegrationTesting before committing.
In fact before I asked Jukka to take a look at this I added
MutableTreeTest#modifiedAfterRebase since I thought it demonstrates the
root cause for testLockWithPendingChanges failing. As it seems the
modifiedAfterRebase test doesn't catch the whole problem since with
Jukka's fix this test passes while testLockWithPendingChanges still fails.
It looks like the root cause of the failure is in the locking
implementation and was just triggered by r1558270, so I'll mark it as
a known issue for now and follow up in OAK-150.
I don't think this is a locking problem since it only occurs with the
SegmentMK.
The test fails since Node.getStatus() returns EXISTING for a modified
node [1]. Not sure why the returned status is still wrong after Jukka's
recent fix though.
Michael
[1]
https://github.com/apache/jackrabbit-oak/blob/trunk/oak-jcr/src/main/java/org/apache/jackrabbit/oak/jcr/lock/LockManagerImpl.java#L146-146
BR,
Jukka Zitting
