[
http://opensource.atlassian.com/projects/hibernate/browse/HHH-6880?page=c...
]
Mindaugas Žakšauskas commented on HHH-6880:
-------------------------------------------
Steve: Just look at the test I wrote for the linked HHH-6822
MZ: The unit test ({{SessionFactorySerializationTest}}) tests whether {{SessionFactory}}
is correctly serialized. That's fine. My point was that {{o.h.t.ComponentType}} (and
others) should not even *attempt* to serialize {{typeScope}} (which is of type
{{o.h.t.TypeFactory.TypeScope}} and which contains reference to {{SessionFactory}}). Why
serialize something what not used when deserialized (in fact, it is +*not used at all*+,
but that's another matter)? In our system we need to replicate billions of Hibernate
cache flush events over the network, and each unnecessary transferred byte just adds to
the delay to the overall system consistency.
Steve: If you don't upgrade major libraries for point releases god have mercy on your
soul...
MZ: 4.0.1 is a bugfix release to 4.0.0. We are running 3.6.7 and do not plan to upgrade to
4.X for a while. Correct me if I'm wrong, but as I understand 4.X is a major upgrade
over 3.X.
Steve: as for the others, I don't think they really require response.
MZ: are you saying that having references to non-serializable objects (comparators) is OK
in a serializable class ({{org.hibernate.type.SortedSetType}})?
Implementations of org.hibernate.type.Type contain references to
non-serializable entities
------------------------------------------------------------------------------------------
Key: HHH-6880
URL:
http://opensource.atlassian.com/projects/hibernate/browse/HHH-6880
Project: Hibernate Core
Issue Type: Bug
Components: core
Affects Versions: 3.6.7
Environment: Seen this on MySQL but I believe this is environment-agnostic
Reporter: Mindaugas Žakšauskas
Assignee: Steve Ebersole
Priority: Critical
Labels: serialization
Time Spent: 59m
This is a spin-off from
-
https://forum.hibernate.org/viewtopic.php?f=1&t=1013687&sid=e62b8...
-
http://stackoverflow.com/q/8404559/7345
org.hibernate.cache.CacheKey is an object which is used for L2 cache notifications - e.g.
to flush a cache on a different cluster node when a certain object has been updated.
The problem comes from org.hibernate.cache.CacheKey carrying the type which (sometimes)
happens to have a reference to objects like org.hibernate.impl.SessionFactoryImpl.
This problem is actually not just theoretic but also causes real pain. First, it takes
much longer to broadcast cache flush events. Secondly, I get this exception during
deserialization of such a CacheKey:
java.lang.NullPointerException
at java.util.concurrent.ConcurrentHashMap.get(ConcurrentHashMap.java:768)
at
org.hibernate.impl.SessionFactoryObjectFactory.getNamedInstance(SessionFactoryObjectFactory.java:159)
at org.hibernate.impl.SessionFactoryImpl.readResolve(SessionFactoryImpl.java:753)
at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:39)
at
sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:25)
at java.lang.reflect.Method.invoke(Method.java:597)
at java.io.ObjectStreamClass.invokeReadResolve(ObjectStreamClass.java:1061)
at java.io.ObjectInputStream.readOrdinaryObject(ObjectInputStream.java:1762)
at java.io.ObjectInputStream.readObject0(ObjectInputStream.java:1329)
at java.io.ObjectInputStream.defaultReadFields(ObjectInputStream.java:1947)
at java.io.ObjectInputStream.readSerialData(ObjectInputStream.java:1871)
at java.io.ObjectInputStream.readOrdinaryObject(ObjectInputStream.java:1753)
at java.io.ObjectInputStream.readObject0(ObjectInputStream.java:1329)
at java.io.ObjectInputStream.defaultReadFields(ObjectInputStream.java:1947)
at java.io.ObjectInputStream.readSerialData(ObjectInputStream.java:1871)
at java.io.ObjectInputStream.readOrdinaryObject(ObjectInputStream.java:1753)
at java.io.ObjectInputStream.readObject0(ObjectInputStream.java:1329)
at java.io.ObjectInputStream.defaultReadFields(ObjectInputStream.java:1947)
at java.io.ObjectInputStream.readSerialData(ObjectInputStream.java:1871)
at java.io.ObjectInputStream.readOrdinaryObject(ObjectInputStream.java:1753)
at java.io.ObjectInputStream.readObject0(ObjectInputStream.java:1329)
at java.io.ObjectInputStream.readUnshared(ObjectInputStream.java:441)
at my.class.deserializeKey(EhcacheUtils.java:35)
I have done a quick scan through all the types, and found these problems related to their
serializability:
1) org.hibernate.type.CollectionType, org.hibernate.type.ComponentType and subclasses of
both - they use attribute "typeScope" (of type TypeFactory.TypeScope) which is
assigned in the constructor and kept in the but seems to be never used. The typeScope in
turn has a reference to the SessionImpl. Can it simply be removed?
2) org.hibernate.type.EntityType (and its subclasses) stores typeScope and only uses it
in the getter, so I guess it should be safe to remove it, too.
3) org.hibernate.type.SortedSetType and org.hibernate.type.SortedMapType keeps a
reference to the comparator, but does not enforce its serializability. It should at least
throw a runtime exception in org.hibernate.mapping.Collection::getComparator if the
comparator is not serializable.
--
This message is automatically generated by JIRA.
For more information on JIRA, see:
http://www.atlassian.com/software/jira