You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@openjpa.apache.org by Pinaki Poddar <pp...@bea.com> on 2008/02/01 19:35:02 UTC
Does EMF need to be Serializable?
Hi,
If serializability is only required because EMF need to be bound to
JNDI context, then it may sense to make EMFImpl implement
javax.naming.Referenceable and no more java.io.Serializable.
This may be the preferred route because EMF configuration is now partly
dynamic. If EMF is bound to JNDI as a serialized instance, then any
dynamic change to its configuration should rebind the EMF to JNDI.
-----Original Message-----
From: David Ezzio [mailto:dezzio@apache.org]
Sent: Friday, February 01, 2008 12:11 PM
To: dev@openjpa.apache.org
Subject: Re: Failures in TeamCity Derby build and svn commit: r616972
Hi,
As to testing that the EMF objects are useful after serialization, this
is a reasonable expectation of any object that is an instance of the
Serializable interface.
If it turns out that we don't want to support serializing useful EMF
objects, then we should take care to remove the Serializable interface
from their implementation.
As I understand it, it is this latter issue that is currently under
discussion.
An important, but only tangentially related, issue is why the build is
so fragile.
David
Pinaki Poddar wrote:
> 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:kwsutter@gmail.com]
> Sent: Friday, February 01, 2008 11:11 AM
> To: dev@openjpa.apache.org
> Subject: Re: Failures in TeamCity Derby build and svn commit: r616972
>
> On Feb 1, 2008 10:17 AM, Pinaki Poddar <pp...@bea.com> 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:plinskey@gmail.com]
>> Sent: Thursday, January 31, 2008 7:57 PM
>> To: dev@openjpa.apache.org
>> 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 <pl...@gmail.com> 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 <de...@apache.org> 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 <pl...@gmail.com>
>> 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, <de...@apache.org> 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.
>
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.