CURATOR-558 - part 1 of ZK 3.6 updates #344
Conversation
| assertEvent(TreeCacheEvent.Type.INITIALIZED, null, null, true); | ||
| assertEvent(TreeCacheEvent.Type.NODE_REMOVED, "/test/me", "data".getBytes(), true); |
There was a problem hiding this comment.
Is this change of order of assertions deliberate (it was previously NODE_REMOVED, then INITIALIZED, it is now INITIALIZED then NODE_REMOVED
There was a problem hiding this comment.
@cammckenzie I moved all kill session calls to use zooKeeper.getTestable().injectSessionExpiration(). When I did that, this particular test's message ordering changed. I didn't think it was too significant. Thoughts?
There was a problem hiding this comment.
Probably not significant, just noticed that the ordering had changed. Interesting that the session injection call has changed the ordering though, given that under the covers it is using the same mechanism.
There was a problem hiding this comment.
The one in ZooKeeperTestable works slightly differently - that's probably why.
There was a problem hiding this comment.
Rightyo, I thought it was calling the same code under the covers. Anyway, I'm OK with it.
There was a problem hiding this comment.
Interesting. The different is that zookeeper's testable has these additional lines
this.clientCnxn.eventThread.queueEventOfDeath();
this.clientCnxn.state = States.CLOSED;
this.clientCnxn.sendThread.getClientCnxnSocket().onClosing();where both of them calls
this.clientCnxn.eventThread.queueEvent(new WatchedEvent(EventType.None, KeeperState.Expired, (String)null));I'm not very sure how it changes the behavior.
There was a problem hiding this comment.
I wrote the ZK testable version but, tbh, I can't remember the specifics.
asfgit
force-pushed
the
CURATOR-558-pt1-remove-zk40-etc
branch
from
e943f7a to
45a6665
Compare
asfgit
force-pushed
the
CURATOR-558-pt1-remove-zk40-etc
branch
2 times, most recently
from
5574e4d to
48434af
Compare
asfgit
force-pushed
the
CURATOR-558-pt1-remove-zk40-etc
branch
from
48434af to
e6424b4
Compare
|
@cammckenzie @shayshim @tisonkun I'm going to merge this unless there is objection. Speak now if you have any concerns. |
| * Upgraded version of ListenerContainer that | ||
| * doesn't leak Guava's internals and also supports mapping/wrapping of listeners |
There was a problem hiding this comment.
| * Upgraded version of ListenerContainer that | |
| * doesn't leak Guava's internals and also supports mapping/wrapping of listeners | |
| * Subclass of {@link ListenerManager} that supports mapping/wrapping of listeners. |
Since ListenerContainer removed, I don't think we have to mention this history in the nightly document.
| @@ -74,7 +74,7 @@ | |||
| private final CloseableExecutorService executorService; | |||
| private final boolean cacheData; | |||
| private final boolean dataIsCompressed; | |||
| private final ListenerContainer<PathChildrenCacheListener> listeners = new ListenerContainer<PathChildrenCacheListener>(); | |||
| private final StandardListenerManager<PathChildrenCacheListener> listeners = StandardListenerManager.standard(); | |||
There was a problem hiding this comment.
Good! I've thought of these changes some days ago.
asfgit
force-pushed
the
CURATOR-558-pt1-remove-zk40-etc
branch
from
e6424b4 to
c4e9432
Compare
Pt1 of change * Remove the ZK 3.4 compatibility module and code * Remove the deprecated ListenerContainer that leaks Guava classes into our APIs * Remove Exhibitor support * Various minor changes/cleanups
asfgit
force-pushed
the
CURATOR-558-pt1-remove-zk40-etc
branch
from
c4e9432 to
3f0f951
Compare
Pt1 of change * Remove the ZK 3.4 compatibility module and code * Remove the deprecated ListenerContainer that leaks Guava classes into our APIs * Remove Exhibitor support * Various minor changes/cleanups Co-authored-by: randgalt <[email protected]>
Pt 1 of change