[hibernate-dev] Problems upgrading ORM 4.3 branch to use Infinispan 7.2.1.Final

Steve Ebersole steve at hibernate.org
Mon May 11 12:14:37 EDT 2015


WRT 4.4, nooooooooooooooooo! :)

On Mon, May 11, 2015 at 10:52 AM, Sanne Grinovero <sanne at hibernate.org>
wrote:

> On 11 May 2015 at 16:35, Galder Zamarreño <galder at redhat.com> wrote:
> >
> >> On 11 May 2015, at 17:26, Sanne Grinovero <sanne at hibernate.org> wrote:
> >>
> >> Any reason to not upgrade Hibernate ORM 4.3 to use Infinispan
> 7.2.1.Final ?
> >> Was Infinispan 6 friendly to JDK 6 users?
> >
> > Yup, Infinispan 6 requires JDK 6, and Infinispan 7 requires JDK7.
> >
> > Also, on top of JDK changes, not sure it'd be a good idea to make a
> dependency update of that magnitude in a bug fix release, since it'd change
> the Infinispan version used, and you'd break all the standalone 4.3 users
> that have tweaked Infinispan configuration, they'd have to move to the more
> WF/AS-like XML configuration. Those using default would be fine since we
> could tweak the default configuration.
>
> Ok, many good reasons :)
>
> But it's a worrying divergence between the catefory of users running
> in WildFly and who doesn't. Also we'd want CI tests for both
> Infinispan integrations, which is not trivial because of all the
> changes in the Infinispan configuration format.
>
> Should this be a 4.4.0 release?
>
> Regarding the JDK, one could say that Java 7 is required when using
> the Infinispan cache. This is something we'd need to document for
> Hibernate ORM 5 anyway, might as well have it apply to a 4.4.
>
> Thanks,
> Sanne
>
> >
> > Cheers,
> >
> >>
> >> On 11 May 2015 at 16:23, Galder Zamarreño <galder at redhat.com> wrote:
> >>>
> >>>> On 11 May 2015, at 16:45, Scott Marlow <smarlow at redhat.com> wrote:
> >>>>
> >>>> Do we need to have a Hibernate ORM continuous integration test setup
> to run against Infinispan 7.2.1?  How about Infinispan 8.x (master?).  Or
> is ORM/Infinispan already tested as part of the Infinispan CI testing?
> >>>
> >>> In this particular case, as mentioned below, running ORM 4.3 testsuite
> with Infinispan 7.2.1 would mostly fail because of XML changes. Again, this
> does not affect WF, so it would not really help you much.
> >>>
> >>> We do have some tests in Infinispan CI to test integration with HB but
> need some updates [1].
> >>>
> >>> Cheers,
> >>>
> >>> [1]
> http://ci.infinispan.org/project.html?projectId=HibernateIntegration
> >>>
> >>>>
> >>>> On 05/11/2015 04:31 AM, Galder Zamarreño wrote:
> >>>>> Hi Gail,
> >>>>>
> >>>>> I've sent a PR for 4.3 to fix HHH-9781 and HHH-9776 in a way that it
> works fine with both Infinispan 7.x and 6.x:
> >>>>> https://github.com/hibernate/hibernate-orm/pull/948
> >>>>>
> >>>>> Cheers,
> >>>>>
> >>>>>> On 11 May 2015, at 09:48, Galder Zamarreño <galder at redhat.com>
> wrote:
> >>>>>>
> >>>>>> Hi Gail,
> >>>>>>
> >>>>>> Thanks for looking into this.
> >>>>>>
> >>>>>> For the scope of WF, the XML part is irrelevant since WF does its
> own configuration parsing, and hence there's no need to make any such
> changes in ORM 4.3. If someone wants to use ORM 4.3 with Infinispan 7.2.x
> standalone, then yes, they need to adjust XML configuration.
> >>>>>>
> >>>>>> The evict/clear issues that 7.2 brought up, and the incorrect
> element count can be fixed in ORM 4.3 without the need to up the dependency
> to 7.2. We just need to apply the changes in a way that work regardless of
> whether 6.0 or 7.2 is used.
> >>>>>>
> >>>>>> I'll work on that today.
> >>>>>>
> >>>>>> Cheers,
> >>>>>>
> >>>>>>> On 11 May 2015, at 07:49, Gail Badner <gbadner at redhat.com> wrote:
> >>>>>>>
> >>>>>>> I ran into some issues upgrading to Infinispan 7.2.1.Final in 4.3
> branch.
> >>>>>>>
> >>>>>>> I cherry-picked the 2 commits for HHH-9632 to upgrade 4.3 to use
> Infinispan 7.1.0.Final:
> >>>>>>>
> >>>>>>> 1)
> https://github.com/hibernate/hibernate-orm/commit/260ff03ae5e8cce0d1d56484e32825222e3046d5
> >>>>>>> 2)
> https://github.com/hibernate/hibernate-orm/commit/1b7e112994413559484e6873f019c5e2c557506b
> >>>>>>> (The 3rd (merge) commit (2cff88cac76147ebb0da5bff8d3605c8a109fd26)
> appeared duplicate 1b7e112994413559484e6873f019c5e2c557506b).
> >>>>>>>
> >>>>>>> 1) updated infinispan-configs.xml to use
> >>>>>>> <infinispan xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
> >>>>>>>      xmlns="urn:infinispan:config:7.0"
> >>>>>>>      xsi:schemaLocation="urn:infinispan:config:7.0
> http://www.infinispan.org/schemas/infinispan-config-7.0.xsd">
> >>>>>>>
> >>>>>>> After 1) was cherry-picked, I was able to build and run tests
> successfully using Infinispan 7.1.0.Final. When I tried running tests
> (without cleaning) using Infinispan 6.0.0.Final, there were lots of test
> failures due to problems configuring the cache:
> >>>>>>>      Caused by: org.hibernate.cache.CacheException
> >>>>>>>          Caused by:
> org.infinispan.commons.CacheConfigurationException
> >>>>>>>              Caused by: javax.xml.stream.XMLStreamException
> >>>>>>>
> >>>>>>> I've attached the output from test
> org.hibernate.test.cache.infinispan.InfinispanRegionFactoryTestCase.
> >>>>>>>
> >>>>>>> Do we need to continue to support running Infinispan 6.0.0.Final
> in ORM 4.3 branch? Could an application have its own dependence on
> Infinispan 6.0.0.Final?
> >>>>>>>
> >>>>>>> Next I cherry-picked the commit for HHH-9781 to upgrade Infinispan
> to 7.2.1.Final:
> https://github.com/hibernate/hibernate-orm/commit/37494f4a9f31c7eaa3486542cb2014b1d3756a87.
> After rebuilding,
> >>>>>>>
> org.hibernate.test.cache.infinispan.timestamp.TimestampsRegionImplTestCase
> fails. I've attached the output for that test as well.
> >>>>>>>
> >>>>>>> Galder, HHH-9781 is still open. Is there more work to do on this?
> Am I missing some other commit that would fix the
> TimestampsRegionImplTestCase failure?
> >>>>>>>
> >>>>>>> This commit (for HHH-9781) includes a change to use
> org.infinispan.commons.util.CloseableIterator, which does not appear to be
> in Infinispan 6.0.0.Final, so this would not be backward-compatible either.
> >>>>>>>
> >>>>>>> Next I cherry-picked the commit for HHH-9776:
> https://github.com/hibernate/hibernate-orm/commit/f8186e10c24a4951785ab43dbaadbec3195df2e5.
> TimestampsRegionImplTestCase still fails; there are no other failures.
> >>>>>>>
> >>>>>>> Galder, HHH-9776 is still open. Is there more work to be done on
> that issue?
> >>>>>>>
> >>>>>>> I've pushed this work to my fork for others to see:
> https://github.com/gbadner/hibernate-core/tree/HHH-9632_HHH-9781_HHH-9776.
> >>>>>>>
> >>>>>>> I've postponed creating a pull request and running the TCK until
> we resolve backward-compatibility requirements and the unit test failure.
> It's OK with me if someone wants to go ahead and run the TCK with what I
> have so far.
> >>>>>>>
> >>>>>>> Regards,
> >>>>>>> Gail
> >>>>>>>
> <TEST-org.hibernate.test.cache.infinispan.timestamp.TimestampsRegionImplTestCase.xml><TEST-org.hibernate.test.cache.infinispan.InfinispanRegionFactoryTestCase.xml>
> >>>>>>
> >>>>>>
> >>>>>> --
> >>>>>> Galder Zamarreño
> >>>>>> galder at redhat.com
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>
> >>>>>
> >>>>> --
> >>>>> Galder Zamarreño
> >>>>> galder at redhat.com
> >>>>>
> >>>>>
> >>>>>
> >>>>>
> >>>
> >>>
> >>> --
> >>> Galder Zamarreño
> >>> galder at redhat.com
> >>>
> >>>
> >>>
> >>>
> >
> >
> > --
> > Galder Zamarreño
> > galder at redhat.com
> >
> >
> >
> >
>


More information about the hibernate-dev mailing list