Question about Jandex composite indexes passed into via "hibernate.jandex_index" and whether ORM keeps a strong reference to Jandex indexes after deployment...
8:40 a.m.
Steve,
Wildfly-dev [1] brings up a change expected for WildFly 10, to not
reference the Jandex indexes after deployment completes. I'm curious if
our current/planned ORM 5.0 jandex using code, keeps a reference to the
passed in "hibernate.jandex_index"?
How how hard would it be for ORM, to only reference the composite Jandex
index during application deployment time? Deployment time ends when
PersistenceProvider.createContainerEntityManagerFactory() returns.
Scott
[1] http://lists.jboss.org/pipermail/wildfly-dev/2015-March/003679.html
10:22 a.m.
I don't *think* we do, but it sounds like something we definitely need to
make sure we aren't doing.
On Mar 17, 2015 8:41 AM, "Scott Marlow" <smarlow(a)redhat.com> wrote:
Steve,
Wildfly-dev [1] brings up a change expected for WildFly 10, to not
reference the Jandex indexes after deployment completes. I'm curious if
our current/planned ORM 5.0 jandex using code, keeps a reference to the
passed in "hibernate.jandex_index"?
How how hard would it be for ORM, to only reference the composite Jandex
index during application deployment time? Deployment time ends when
PersistenceProvider.createContainerEntityManagerFactory() returns.
Scott
[1] http://lists.jboss.org/pipermail/wildfly-dev/2015-March/003679.html
_______________________________________________
hibernate-dev mailing list
hibernate-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/hibernate-dev
12:05 p.m.
Well I should also clarify that currently we are not using Jandex. That
was all work done as part of the mapping/metamodel redesign which we
decided to push to 6.0.
On Wed, Mar 18, 2015 at 10:22 AM, Steve Ebersole <steve(a)hibernate.org>
wrote:
I don't *think* we do, but it sounds like something we definitely
need to
make sure we aren't doing.
On Mar 17, 2015 8:41 AM, "Scott Marlow" <smarlow(a)redhat.com> wrote:
> Steve,
>
> Wildfly-dev [1] brings up a change expected for WildFly 10, to not
> reference the Jandex indexes after deployment completes. I'm curious if
> our current/planned ORM 5.0 jandex using code, keeps a reference to the
> passed in "hibernate.jandex_index"?
>
> How how hard would it be for ORM, to only reference the composite Jandex
> index during application deployment time? Deployment time ends when
> PersistenceProvider.createContainerEntityManagerFactory() returns.
>
> Scott
>
> [1] http://lists.jboss.org/pipermail/wildfly-dev/2015-March/003679.html
> _______________________________________________
> hibernate-dev mailing list
> hibernate-dev(a)lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/hibernate-dev
>
12:44 p.m.
On 03/18/2015 01:05 PM, Steve Ebersole wrote:
Well I should also clarify that currently we are not using Jandex.
That
was all work done as part of the mapping/metamodel redesign which we
decided to push to 6.0.
I assume you mean the master (5.0) code base doesn't use Jandex but it
looks like [2] in 4.3, is building/using the Jandex index.
How big of a change would it be to use Jandex (via
"hibernate.jandex_index") in ORM 4.3/5.0 mapping? Would be nice to have
a comparison of how long it takes to deploy a large application with and
without Jandex.
Scott
[2]
https://github.com/hibernate/hibernate-orm/blob/4.3/hibernate-entitymanag...
On Wed, Mar 18, 2015 at 10:22 AM, Steve Ebersole <steve(a)hibernate.org
<mailto:steve@hibernate.org>> wrote:
I don't *think* we do, but it sounds like something we definitely
need to make sure we aren't doing.
On Mar 17, 2015 8:41 AM, "Scott Marlow" <smarlow(a)redhat.com
<mailto:smarlow@redhat.com>> wrote:
Steve,
Wildfly-dev [1] brings up a change expected for WildFly 10, to not
reference the Jandex indexes after deployment completes. I'm
curious if
our current/planned ORM 5.0 jandex using code, keeps a reference
to the
passed in "hibernate.jandex_index"?
How how hard would it be for ORM, to only reference the
composite Jandex
index during application deployment time? Deployment time ends when
PersistenceProvider.createContainerEntityManagerFactory() returns.
Scott
[1]
http://lists.jboss.org/pipermail/wildfly-dev/2015-March/003679.html
_______________________________________________
hibernate-dev mailing list
hibernate-dev(a)lists.jboss.org <mailto:hibernate-dev@lists.jboss.org>
https://lists.jboss.org/mailman/listinfo/hibernate-dev
12:49 p.m.
Massive.
On Wed, Mar 18, 2015 at 12:44 PM, Scott Marlow <smarlow(a)redhat.com> wrote:
On 03/18/2015 01:05 PM, Steve Ebersole wrote:
> Well I should also clarify that currently we are not using Jandex. That
> was all work done as part of the mapping/metamodel redesign which we
> decided to push to 6.0.
>
I assume you mean the master (5.0) code base doesn't use Jandex but it
looks like [2] in 4.3, is building/using the Jandex index.
How big of a change would it be to use Jandex (via
"hibernate.jandex_index") in ORM 4.3/5.0 mapping? Would be nice to have a
comparison of how long it takes to deploy a large application with and
without Jandex.
Scott
[2] https://github.com/hibernate/hibernate-orm/blob/4.3/
hibernate-entitymanager/src/main/java/org/hibernate/jpa/boot/internal/
EntityManagerFactoryBuilderImpl.java
> On Wed, Mar 18, 2015 at 10:22 AM, Steve Ebersole <steve(a)hibernate.org
> <mailto:steve@hibernate.org>> wrote:
>
> I don't *think* we do, but it sounds like something we definitely
> need to make sure we aren't doing.
>
> On Mar 17, 2015 8:41 AM, "Scott Marlow" <smarlow(a)redhat.com
> <mailto:smarlow@redhat.com>> wrote:
>
> Steve,
>
> Wildfly-dev [1] brings up a change expected for WildFly 10, to not
> reference the Jandex indexes after deployment completes. I'm
> curious if
> our current/planned ORM 5.0 jandex using code, keeps a reference
> to the
> passed in "hibernate.jandex_index"?
>
> How how hard would it be for ORM, to only reference the
> composite Jandex
> index during application deployment time? Deployment time ends
> when
> PersistenceProvider.createContainerEntityManagerFactory()
> returns.
>
> Scott
>
> [1]
> http://lists.jboss.org/pipermail/wildfly-dev/2015-
> March/003679.html
> _______________________________________________
> hibernate-dev mailing list
> hibernate-dev(a)lists.jboss.org <mailto:hibernate-dev@lists.
> jboss.org>
> https://lists.jboss.org/mailman/listinfo/hibernate-dev
>
>
>
12:53 p.m.
Out of curiosity though, since you want more and more and more things
pushed to "second phase bootstrapping", I am curious how this new
expectation plays in with that. Specifically, us "using" Jandex will
certainly include us accessing classes, unless Jason has added the expanded
capabilities he and I have discussed to Jandex wrt accessing Class members
outside of the annotations they may or may not have.
On Wed, Mar 18, 2015 at 12:49 PM, Steve Ebersole <steve(a)hibernate.org>
wrote:
Massive.
On Wed, Mar 18, 2015 at 12:44 PM, Scott Marlow <smarlow(a)redhat.com> wrote:
> On 03/18/2015 01:05 PM, Steve Ebersole wrote:
>
>> Well I should also clarify that currently we are not using Jandex. That
>> was all work done as part of the mapping/metamodel redesign which we
>> decided to push to 6.0.
>>
>
> I assume you mean the master (5.0) code base doesn't use Jandex but it
> looks like [2] in 4.3, is building/using the Jandex index.
>
> How big of a change would it be to use Jandex (via
> "hibernate.jandex_index") in ORM 4.3/5.0 mapping? Would be nice to have a
> comparison of how long it takes to deploy a large application with and
> without Jandex.
>
> Scott
>
> [2] https://github.com/hibernate/hibernate-orm/blob/4.3/
> hibernate-entitymanager/src/main/java/org/hibernate/jpa/boot/internal/
> EntityManagerFactoryBuilderImpl.java
>
>
>> On Wed, Mar 18, 2015 at 10:22 AM, Steve Ebersole <steve(a)hibernate.org
>> <mailto:steve@hibernate.org>> wrote:
>>
>> I don't *think* we do, but it sounds like something we definitely
>> need to make sure we aren't doing.
>>
>> On Mar 17, 2015 8:41 AM, "Scott Marlow" <smarlow(a)redhat.com
>> <mailto:smarlow@redhat.com>> wrote:
>>
>> Steve,
>>
>> Wildfly-dev [1] brings up a change expected for WildFly 10, to
>> not
>> reference the Jandex indexes after deployment completes. I'm
>> curious if
>> our current/planned ORM 5.0 jandex using code, keeps a reference
>> to the
>> passed in "hibernate.jandex_index"?
>>
>> How how hard would it be for ORM, to only reference the
>> composite Jandex
>> index during application deployment time? Deployment time ends
>> when
>> PersistenceProvider.createContainerEntityManagerFactory()
>> returns.
>>
>> Scott
>>
>> [1]
>> http://lists.jboss.org/pipermail/wildfly-dev/2015-
>> March/003679.html
>> _______________________________________________
>> hibernate-dev mailing list
>> hibernate-dev(a)lists.jboss.org <mailto:hibernate-dev@lists.
>> jboss.org>
>> https://lists.jboss.org/mailman/listinfo/hibernate-dev
>>
>>
>>
1:56 p.m.
On 03/18/2015 01:53 PM, Steve Ebersole wrote:
Out of curiosity though, since you want more and more and more
things
pushed to "second phase bootstrapping", I am curious how this new
expectation plays in with that. Specifically, us "using" Jandex will
certainly include us accessing classes, unless Jason has added the
expanded capabilities he and I have discussed to Jandex wrt accessing
Class members outside of the annotations they may or may not have.
If we use the application classloader during the "first phase
bootstrapping", entity class enhancement will not work (application
classes will be loaded before they can be rewritten). Classloading
during the "first phase bootstrapping" with
javax.persistence.spi.PersistenceUnitInfo.getNewTempClassLoader() would
still work with entity class enhancement.
On Wed, Mar 18, 2015 at 12:49 PM, Steve Ebersole <steve(a)hibernate.org
<mailto:steve@hibernate.org>> wrote:
Massive.
On Wed, Mar 18, 2015 at 12:44 PM, Scott Marlow <smarlow(a)redhat.com
<mailto:smarlow@redhat.com>> wrote:
On 03/18/2015 01:05 PM, Steve Ebersole wrote:
Well I should also clarify that currently we are not using
Jandex. That
was all work done as part of the mapping/metamodel redesign
which we
decided to push to 6.0.
I assume you mean the master (5.0) code base doesn't use Jandex
but it looks like [2] in 4.3, is building/using the Jandex index.
How big of a change would it be to use Jandex (via
"hibernate.jandex_index") in ORM 4.3/5.0 mapping? Would be nice
to have a comparison of how long it takes to deploy a large
application with and without Jandex.
Scott
[2]
https://github.com/hibernate/__hibernate-orm/blob/4.3/__hibernate-entitym...
<https://github.com/hibernate/hibernate-orm/blob/4.3/hibernate-entitymanag...
On Wed, Mar 18, 2015 at 10:22 AM, Steve Ebersole
<steve(a)hibernate.org <mailto:steve@hibernate.org>
<mailto:steve@hibernate.org <mailto:steve@hibernate.org>>>
wrote:
I don't *think* we do, but it sounds like something we
definitely
need to make sure we aren't doing.
On Mar 17, 2015 8:41 AM, "Scott Marlow"
<smarlow(a)redhat.com <mailto:smarlow@redhat.com>
<mailto:smarlow@redhat.com
<mailto:smarlow@redhat.com>>> wrote:
Steve,
Wildfly-dev [1] brings up a change expected for
WildFly 10, to not
reference the Jandex indexes after deployment
completes. I'm
curious if
our current/planned ORM 5.0 jandex using code,
keeps a reference
to the
passed in "hibernate.jandex_index"?
How how hard would it be for ORM, to only reference the
composite Jandex
index during application deployment time?
Deployment time ends when
PersistenceProvider.__createContainerEntityManagerFa__ctory() returns.
Scott
[1]
http://lists.jboss.org/__pipermail/wildfly-dev/2015-__March/003679.html
<http://lists.jboss.org/pipermail/wildfly-dev/2015-March/003679.html>
_________________________________________________
hibernate-dev mailing list
hibernate-dev(a)lists.jboss.org
<mailto:hibernate-dev@lists.jboss.org>
<mailto:hibernate-dev@lists.__jboss.org
<mailto:hibernate-dev@lists.jboss.org>>
https://lists.jboss.org/__mailman/listinfo/hibernate-dev
<https://lists.jboss.org/mailman/listinfo/hibernate-dev>
2:24 p.m.
On 03/18/2015 02:56 PM, Scott Marlow wrote:
On 03/18/2015 01:53 PM, Steve Ebersole wrote:
> Out of curiosity though, since you want more and more and more things
> pushed to "second phase bootstrapping", I am curious how this new
> expectation plays in with that. Specifically, us "using" Jandex will
> certainly include us accessing classes, unless Jason has added the
> expanded capabilities he and I have discussed to Jandex wrt accessing
> Class members outside of the annotations they may or may not have.
If we use the application classloader during the "first phase
bootstrapping", entity class enhancement will not work (application
classes will be loaded before they can be rewritten). Classloading
during the "first phase bootstrapping" with
javax.persistence.spi.PersistenceUnitInfo.getNewTempClassLoader() would
still work with entity class enhancement.
Jipijapa for ORM 4.1 is currently using the
PersistenceUnitInfo.getNewTempClassLoader() [3] and that works well.
Jason asked why we dropped using Jandex in our integration with ORM 4.3.
The 4.3 integration in Jipijapa is more elegant/better except it might
not be as fast as using Jandex with
org.hibernate.boot.archive.scan.spi.Scanner.
Should we look at switching Jipijapa back to
org.hibernate.boot.archive.scan.spi.Scanner (so we can use Jandex) and
remove Jandex from 4.3.x?
In the Jipijapa integration for ORM 5.0, should we use
org.hibernate.boot.archive.scan.spi.Scanner so that we are using jandex?
[3]
https://github.com/jipijapa/jipijapa/blob/master/hibernate4_1/src/main/ja...
>
> On Wed, Mar 18, 2015 at 12:49 PM, Steve Ebersole <steve(a)hibernate.org
> <mailto:steve@hibernate.org>> wrote:
>
> Massive.
>
> On Wed, Mar 18, 2015 at 12:44 PM, Scott Marlow <smarlow(a)redhat.com
> <mailto:smarlow@redhat.com>> wrote:
>
> On 03/18/2015 01:05 PM, Steve Ebersole wrote:
>
> Well I should also clarify that currently we are not using
> Jandex. That
> was all work done as part of the mapping/metamodel redesign
> which we
> decided to push to 6.0.
>
>
> I assume you mean the master (5.0) code base doesn't use Jandex
> but it looks like [2] in 4.3, is building/using the Jandex index.
>
> How big of a change would it be to use Jandex (via
> "hibernate.jandex_index") in ORM 4.3/5.0 mapping? Would be nice
> to have a comparison of how long it takes to deploy a large
> application with and without Jandex.
>
> Scott
>
> [2]
>
https://github.com/hibernate/__hibernate-orm/blob/4.3/__hibernate-entitym...
>
<https://github.com/hibernate/hibernate-orm/blob/4.3/hibernate-entitymanag...
>
>
> On Wed, Mar 18, 2015 at 10:22 AM, Steve Ebersole
> <steve(a)hibernate.org <mailto:steve@hibernate.org>
> <mailto:steve@hibernate.org
<mailto:steve@hibernate.org>>>
> wrote:
>
> I don't *think* we do, but it sounds like something we
> definitely
> need to make sure we aren't doing.
>
> On Mar 17, 2015 8:41 AM, "Scott Marlow"
> <smarlow(a)redhat.com <mailto:smarlow@redhat.com>
> <mailto:smarlow@redhat.com
> <mailto:smarlow@redhat.com>>> wrote:
>
> Steve,
>
> Wildfly-dev [1] brings up a change expected for
> WildFly 10, to not
> reference the Jandex indexes after deployment
> completes. I'm
> curious if
> our current/planned ORM 5.0 jandex using code,
> keeps a reference
> to the
> passed in "hibernate.jandex_index"?
>
> How how hard would it be for ORM, to only reference the
> composite Jandex
> index during application deployment time?
> Deployment time ends when
>
> PersistenceProvider.__createContainerEntityManagerFa__ctory() returns.
>
> Scott
>
> [1]
> http://lists.jboss.org/__pipermail/wildfly-dev/2015-__March/003679.html
>
<http://lists.jboss.org/pipermail/wildfly-dev/2015-March/003679.html>
> _________________________________________________
> hibernate-dev mailing list
> hibernate-dev(a)lists.jboss.org
> <mailto:hibernate-dev@lists.jboss.org>
> <mailto:hibernate-dev@lists.__jboss.org
> <mailto:hibernate-dev@lists.jboss.org>>
> https://lists.jboss.org/__mailman/listinfo/hibernate-dev
> <https://lists.jboss.org/mailman/listinfo/hibernate-dev>
>
>
>
>
_______________________________________________
hibernate-dev mailing list
hibernate-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/hibernate-dev
3341
days inactive
3342
days old
7 comments
2 participants
participants (2)
-
Scott Marlow -
Steve Ebersole