Mailing-List: contact dev-help@openjpa.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@openjpa.apache.org
Message-ID: <32980043.316011285077333490.JavaMail.jira@thor>
Date: Tue, 21 Sep 2010 09:55:33 -0400 (EDT)
From: "Willis Blackburn (JIRA)" <jira@apache.org>
To: dev@openjpa.apache.org
Subject: [jira] Commented: (OPENJPA-1804) NPE in MappingInfo.java line 1514
In-Reply-To: <7307874.276561284839673337.JavaMail.jira@thor>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable


    [ https://issues.apache.org/jira/browse/OPENJPA-1804?page=3Dcom.atlassi=
an.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=3D12=
912984#action_12912984 ]=20

Willis Blackburn commented on OPENJPA-1804:
-------------------------------------------

Rick, you asked, "Perhaps we need to have a stronger warning message when t=
his feature is enabled? ... or disable it all together?"

I saw on the roadmap that OpenJPA 2.1 will require Java 6.  I think that if=
 OpenJPA is not going to support Java 6, then there is no need to support w=
orking with unenhanced classes.  The strategy of using the attach API to in=
stall the agent will work for casual users, and the others can pass the age=
nt on the command line, use the J2EE container hooks, or use the build-time=
 enhancer.  The behavior of OpenJPA when using unenhanced classes is differ=
ent from its behavior when using enhanced classes (no lazy loading, changes=
 only detected at flush), so anything that users actually develop in that m=
ode is not going to be production-ready.  At some point developers will hav=
e to figure out how to get the agent or the build-time enhancer working and=
 test their code under that configuration.  Given how easy it is to use the=
 attach API enhancer under Java 6, there's not a strong case for having a l=
ess-functional alternative.

Have I added enough comments to this issue yet?  I think I'm going for a pe=
rsonal record.

> NPE in MappingInfo.java line 1514
> ---------------------------------
>
>                 Key: OPENJPA-1804
>                 URL: https://issues.apache.org/jira/browse/OPENJPA-1804
>             Project: OpenJPA
>          Issue Type: Bug
>          Components: Enhance, UnenhancedClasses, usability
>    Affects Versions: 2.0.1
>            Reporter: Willis Blackburn
>
> I am encountering a NPE at MappingInfo.java line 1514.  I'm sorry I don't=
 have time to prepare a test case right now.  I'm not even 100% sure what i=
s causing the problem, but it seems to be something like this:
> I have classes Organization, Group, User, and AddressBook.
> Group has a reference (many-to-one) to Organization.
> I'm using field-level access and the Java 6 runtime enhancer.  (NOT the b=
uild-time enhancer or the agent.)
> In ManagedClassSubclasser.prepareUnenhancedClasses, the code receives a l=
ist of Class objects--the ones in persistence.xml.  They are not in the sam=
e order as they are in persistence.xml, but I don't know if that's importan=
t.  The first one is Group.  During the processing of Group, the code invok=
es ClassMapping.setTable to set the table name, which is "group_table."  Th=
is comes from the @Table annotation of Group.  So far, so good.  Then the c=
ode does some additional processing on the relationship mappings of Group, =
which include the reference to Organization.  During the processing of that=
 relationship, the code calls MappingInfo.mergeJoinColumn.  On line 1367, t=
he method calls rel.getTable.  The "rel" variable points to the ClassMappin=
g for Organization.  Remember that prepareUnenhancedClasses hasn't gotten t=
o Organization yet--it's still working on Group--and so I assume that this =
Organization mapping was created on-demand.  However, rel.getTable returns =
null.  The Organization class has a @Table attribute, and I can see that th=
e table name has been loaded in the ClassMappingInfo instance attached to t=
he ClassMapping for Organization, however the table name has not yet been c=
opied from ClassMappingInfo to ClassMapping.  Later, the mergeJoinColumn me=
thod attempts to dereference the null table name and generates the NPE.
> Bottom line:  When using the Java 6, runtime enhancer, if class A has a r=
eference to class B, and both classes have @Table annotations (at least), a=
nd class A is processed first, then it produces a NPE.

--=20
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.