Hi,
Serializablity requirement on EMF is arising out of necessity rather
than by JPA specifcation.
JNDI, too, does not mandate bound objects to be Serializable.
But JNDI providers often place that restriction.
Non-serializable objects can be bound to JNDI by implementing
javax.naming.Referenceable.
Kodo's JDO product derivation based on OpenJPA is following this
Referenceable route.
But I do not see this in openjpa EntityManagerFactory implementation.
-----Original Message-----
From: Kevin Sutter [mailto:[EMAIL PROTECTED]
Sent: Friday, February 01, 2008 11:11 AM
To: [email protected]
Subject: Re: Failures in TeamCity Derby build and svn commit: r616972
On Feb 1, 2008 10:17 AM, Pinaki Poddar <[EMAIL PROTECTED]> wrote:
Factories needs to be Serializable as they are often placed in JNDI.
The JNDI lookup needs to work for the Factory definition, but it doesn't
require that the actual Factory be present in JNDI. And, since the JNDI
lookup is meant for the EE (Container) environment, I would guess that
most implementations provide a wrapper object in JNDI for the actual
Factory.
Granted, this wrapper approach is not a requirement. But, I can't find
the requirement in the spec that the EMF has to be Serializable
either... :-)
Kevin
-----Original Message-----
From: Patrick Linskey [mailto:[EMAIL PROTECTED]
Sent: Thursday, January 31, 2008 7:57 PM
To: [email protected]
Subject: Re: Failures in TeamCity Derby build and svn commit: r616972
To be honest, I don't really understand why a BrokerFactory needs to
be serializable, so I think probably it'd be good to make sure that
there is good reason for that before going too deep into this
debugging exercise.
-Patrick
On Jan 31, 2008 5:52 PM, Patrick Linskey <[EMAIL PROTECTED]> wrote:
If the hangup is intolerable, let me know, and I'll back out my
change tomorrow.
Personally, I'd rather have the TeamCity build be clean while you
investigate. As you noted, the error-reporting from surefire is
frustratingly inaccurate, and this really erodes the benefit of
continuous integration.
It would be handy if I knew a few things. How to pass in VM debug
args to Maven for its test runs, where to set the OpenJPA logging
when Maven runs the tests, and how to pass in the appropriate
parameters when running the test case outside of Maven. Any
insights on these questions will be appreciated.
I usually just add log configuration to the test cases themselves,
in the setUp(...) method:
setUp(Class.class, Class2.class, ..., "openjpa.Log",
"SQL=TRACE");
I believe that it is also possible to set this at the command line
via
-Dopenjpa.Log=SQL=TRACE types of clauses.
-Patrick
On Jan 31, 2008 4:08 PM, David Ezzio <[EMAIL PROTECTED]> wrote:
Hi,
Progress has been slow on the test failures apparently introduced
by
616972. I have made progress in reproducing the results.
The miscounting in TeamCity is gross. Although it lists 33
failing tests complete with a credible stack trace for each, in
fact, there is only one failing test.
Meanwhile, I believe now, that change 616658 is no longer a
suspect.
I get very erratic results here, but they seem to vary more by
whether I run "mvn test" after "mvn clean" or "mvn install" after
"mvn clean". By doing just what TeamCity does, "mvn clean compile
test", I can finally reproduce the results locally that TeamCity
reports in its Compile-Derby build for this change (395, I think).
In spite of this, if I run "mvn
-Dtest=TestCaseInsensitiveKeywordsInJPQL clean compile test" the
test passes.
It would be handy if I knew a few things. How to pass in VM debug
args to Maven for its test runs, where to set the OpenJPA logging
when Maven runs the tests, and how to pass in the appropriate
parameters when running the test case outside of Maven. Any
insights on these questions will be appreciated.
But at least at this point, I can reproduce locally what TeamCity
reports, and although it happened only once out of one try, I am
optimistic that it is reproducible. I will continue working this
tomorrow.
If the hangup is intolerable, let me know, and I'll back out my
change tomorrow.
David Ezzio
David Ezzio wrote:
Hi,
I am convinced that change 616658 (small change to
DBDictionary.java) is the source of the two intermittent
failures
that I am seeing locally.
One of those intermittent failures is the failure in
TestCaseInsensitiveKeywordsInJPQL.testCaseInsensitiveBooleans
which is one of the 33 related failures currently seen in
TeamCity
Derby builds.
The other intermittent failure is
TestVersion.testVersionTimestamp
which I reported earlier as occurring locally with change
616658.
When I revert change 616658 locally, both of these issues go
away even with the changes that I submitted to
AbstractBrokerFactory in
change 616972.
I have a question: is it possible to run a build in TeamCity
without submitting the change (or conditionally submitting the
change) to SVN?
I'd like to find out if reverting 616658 will fix the issue. If
so, then I can try to figure out why it's a problem.
Thanks,
David
David Ezzio wrote:
No, but I'm taking a look since my changes appear to be the
cause.
Patrick Linskey wrote:
5. The tests are now failing in the automated build; any idea
why this might be?
-Patrick
On Jan 30, 2008 6:11 PM, Patrick Linskey <[EMAIL PROTECTED]>
wrote:
Hi,
A few questions:
1. Were you seeing this cause problems somewhere? I.e., when
do
we serialize BrokerFactory instances?
2. Why not just move the initialization code to the
declaration
line (instead of the constructor)?
3. It looks like there are other transactional structures
that might not be properly handled in serialization /
deserialization (lifecycle listeners, for example). Should we
be making more changes there?
-Patrick
On Jan 30, 2008 4:59 PM, <[EMAIL PROTECTED]> wrote:
Author: dezzio
Date: Wed Jan 30 16:59:02 2008 New Revision: 616972
URL: http://svn.apache.org/viewvc?rev=616972&view=rev
Log:
Allow EntityManagerFactory objects to be serialized and
deserialized successfully.
Added:
openjpa/trunk/openjpa-persistence-jdbc/src/test/java/org/apa
ch e/openjpa/persistence/simple/TestSerializedFactory.java
(with props)
Modified:
openjpa/trunk/openjpa-kernel/src/main/java/org/apache/openjp
a/ kernel/AbstractBrokerFactory.java
Modified:
openjpa/trunk/openjpa-kernel/src/main/java/org/apache/openjp
a/ kernel/AbstractBrokerFactory.java
URL:
http://svn.apache.org/viewvc/openjpa/trunk/openjpa-kernel/sr
c/
main/java/org/apache/openjpa/kernel/AbstractBrokerFactory.ja
va ?rev=616972&r1=616971&r2=616972&view=diff
============================================================
==
================
---
openjpa/trunk/openjpa-kernel/src/main/java/org/apache/openjp
a/ kernel/AbstractBrokerFactory.java
(original)
+++
openjpa/trunk/openjpa-kernel/src/main/java/org/apache/openjp
a/ kernel/AbstractBrokerFactory.java Wed Jan 30 16:59:02
2008 @@ -147,8 +147,7 @@
*/
protected AbstractBrokerFactory(OpenJPAConfiguration
config) {
_conf = config;
- _pcClassLoaders = new ConcurrentReferenceHashSet(
- ConcurrentReferenceHashSet.WEAK);
+ getPcClassLoaders();
}
/**
@@ -287,13 +286,13 @@
if (needsSub(cls))
toRedefine.add(cls);
}
- _pcClassLoaders.add(loader);
+ getPcClassLoaders().add(loader);
_pcClassNames = c;
}
_persistentTypesLoaded = true;
} else {
// reload with this loader
- if (_pcClassLoaders.add(loader)) {
+ if (getPcClassLoaders().add(loader)) {
for (Iterator itr =
_pcClassNames.iterator();
itr.hasNext();) {
try {
Class cls = @@ -818,4 +817,15 @@
_transactional.remove (_trans);
}
}
+
+ /**
+ * Method insures that deserialized EMF has this
+ reference
re-instantiated
+ */
+ private Collection getPcClassLoaders() {
+ if (_pcClassLoaders == null)
+ _pcClassLoaders = new ConcurrentReferenceHashSet(
+ ConcurrentReferenceHashSet.WEAK);
+
+ return _pcClassLoaders;
+ }
}
Added:
openjpa/trunk/openjpa-persistence-jdbc/src/test/java/org/apa
ch e/openjpa/persistence/simple/TestSerializedFactory.java
URL:
http://svn.apache.org/viewvc/openjpa/trunk/openjpa-persisten
ce
-jdbc/src/test/java/org/apache/openjpa/persistence/simple/Te
st SerializedFactory.java?rev=616972&view=auto
============================================================
==
================
---
openjpa/trunk/openjpa-persistence-jdbc/src/test/java/org/apa
ch e/openjpa/persistence/simple/TestSerializedFactory.java
(added)
+++
openjpa/trunk/openjpa-persistence-jdbc/src/test/java/org/apa
ch e/openjpa/persistence/simple/TestSerializedFactory.java
Wed Jan 30 16:59:02 2008
@@ -0,0 +1,74 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under
+one
+ * or more contributor license agreements. See the NOTICE
+file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this
+file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in
+compliance
+ * with the License. You may obtain a copy of the License
+at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in
+writing,
+ * software distributed under the License is distributed on
+an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied. See the License for
+the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+package org.apache.openjpa.persistence.simple;
+
+import java.io.*;
+
+import javax.persistence.EntityManager; import
+javax.persistence.EntityManagerFactory;
+import javax.persistence.EntityTransaction;
+
+import junit.textui.TestRunner; import
+org.apache.openjpa.persistence.OpenJPAEntityManager;
+import
+org.apache.openjpa.persistence.test.SingleEMFTestCase;
+
+/**
+ * Tests that a EntityManagerFactory can be used after
serialization.
+ *
+ * @author David Ezzio
+ */
+public class TestSerializedFactory
+ extends SingleEMFTestCase {
+
+ public void setUp() {
+ setUp(AllFieldTypes.class);
+ }
+
+ public void testSerializedEntityManagerFactory() throws
Exception {
+ // serialize and deserialize the entity manager
factory
+ ByteArrayOutputStream baos = new
ByteArrayOutputStream();
+ ObjectOutputStream oos = new
ObjectOutputStream(baos);
+ oos.writeObject(emf);
+ EntityManagerFactory emf2 =
+ (EntityManagerFactory) new ObjectInputStream(
+ new
ByteArrayInputStream(baos.toByteArray())).readObject();
+
+ // use the deserialized entity manager factory
+ assertTrue("The deserialized entity manager factory
+ is not
open",
+ emf2.isOpen());
+ EntityManager em = emf2.createEntityManager();
+ assertTrue("The newly created entity manager is not
+ open",
em.isOpen());
+
+ // exercise the entity manager produced from the
deserialized EMF
+ em.getTransaction().begin();
+ em.persist(new AllFieldTypes());
+ em.getTransaction().commit();
+
+ // close the extra resources
+ em.close();
+ assertFalse("The entity manager is not closed",
em.isOpen());
+ emf2.close();
+ assertFalse("The entity manager factory is not
+ closed",
emf2.isOpen());
+ }
+
+ public static void main(String[] args) {
+ TestRunner.run(TestSerializedFactory.class);
+ }
+}
+
Propchange:
openjpa/trunk/openjpa-persistence-jdbc/src/test/java/org/apa
ch e/openjpa/persistence/simple/TestSerializedFactory.java
------------------------------------------------------------
--
----------------
svn:eol-style = native
--
Patrick Linskey
202 669 5907
--
Patrick Linskey
202 669 5907
--
Patrick Linskey
202 669 5907
Notice: This email message, together with any attachments, may
contain information of BEA Systems, Inc., its subsidiaries and
affiliated entities, that may be confidential, proprietary,
copyrighted and/or legally privileged, and is intended solely for the
use of the individual or entity named in this message. If you are not
the intended recipient, and have received this message in error,
please immediately return this by email and then delete it.
Notice: This email message, together with any attachments, may contain
information of BEA Systems, Inc., its subsidiaries and affiliated
entities, that may be confidential, proprietary, copyrighted and/or legally
privileged, and is intended solely for the use of the individual or entity
named in this message. If you are not the intended recipient, and have received
this message in error, please immediately return this by email and then delete
it.
