[hibernate-dev] Bytecode enhanced, Reference Cached immutable Entities
Sanne Grinovero
sanne at hibernate.org
Fri Mar 27 08:14:15 EDT 2015
Also from OGM and Search perspective, we're not expecting any of this
to be backported.
It's not a functional blocker, "just" significant performance improvements.
How do we move further? John, I'm not familiar with the requirements
for the bytecode instrumentation, but can help a bit on the other
aspects. I could either work on extracting the patch of converting the
EntityEntry into an interface from your draft, or if you have time to
polish that I could review that PR. After that change is integrated,
it should be easier for us to work in parallel on different areas and
draft some ideas with the factory context idea.
+1 to an EntityPersister:EntityEntryFactory 1:1 relation, glad to see
it makes sense in your minds as well.
To make a decision on the Immutable entities vs Read Only entities..
are we good in applying the optimisation only to Immutable entities or
do we need to find a way for entities which happen to be loaded
read-only too? I might have an idea to play with for that, but it
obviously gets a bit more complex.
@Steve, I guess we're looking at you for advice on what needs to be
done for the Lock state. I'm understanding that needs to be addressed
right?
Thanks,
Sanne
On 27 March 2015 at 11:42, John O'Hara <johara at redhat.com> wrote:
> Steve,
>
> From our point of view, we would not be looking to backport any features
> introduced to any of the ORM4 releases. Our focus has shifted to the next
> major release of EAP as we have a number of issues we need to overcome.
> Getting this feature into ORM5 sooner rather than later is essential for our
> use case in the future.
>
> Thanks
>
> John
>
>
> On 27/03/15 11:36, Steve Ebersole wrote:
>
> BTW, if we go this route this *cannot* be backported because we do allow
> users to override EntityPersister; its an extension point. Adding a new
> method to extension points in bugfixes is a no-no.
>
> On Fri, Mar 27, 2015 at 6:35 AM, Steve Ebersole <steve at hibernate.org> wrote:
>>
>> An EntityPersister describes each entity you defined in your model. Its
>> one-to-one between @Entity and EntityPersister. And really, it is totally
>> reasonable to have the EntityPersister determine the "EntityEntry builder"
>> to use. That also eliminates needing to develop yet-another-extension-point
>> because if we expose it from EntityPersister OGM already swaps
>> EntityPersister impls so this would JustWork.
>>
>> On Fri, Mar 27, 2015 at 6:04 AM, John O'Hara <johara at redhat.com> wrote:
>>>
>>> On 26/03/15 16:22, Steve Ebersole wrote:
>>>
>>> The contract I suggested btw is missing one use case I know we have
>>> discussed in regards to OGM a few times... the ability to store references
>>> to datastore specific locators (I think the use case was to efficiently load
>>> collections through that datastore reference). But that (obviously)
>>> requires passing extra, specific state into the createEntityEntry method.
>>> Not really sure the best way to handle such things tbh.
>>>
>>> If we did have a "creation Context" parameter passed to the
>>> createEntityEntry, we would be instantiating more objects, however if we
>>> went with Sanne' suggestion if caching the EnityEntryFactory with the
>>> EntityPersister, how many objects would be actually created? I am not sure
>>> about the relationship between Entity, EntityEntry and Session but could
>>> this a possibility if there is a 1-1 relationship between EntityPersister
>>> and no. of different Entities definitions?
>>>
>>>
>>> On Thu, Mar 26, 2015 at 11:16 AM, Steve Ebersole <steve at hibernate.org>
>>> wrote:
>>>>
>>>> Apparently I hit some key combo that means send :) To finish up...
>>>>
>>>> On Thu, Mar 26, 2015 at 10:58 AM, Steve Ebersole <steve at hibernate.org>
>>>> wrote:
>>>>>
>>>>>
>>>>> On Thu, Mar 26, 2015 at 10:20 AM, Sanne Grinovero <sanne at hibernate.org>
>>>>> wrote:
>>>>>>
>>>>>>
>>>>>> Concurrency
>>>>>> Since the goal is to share the ImmutableEntityEntry instance among
>>>>>> multiple threads reading it, I'd rather make sure that it is really
>>>>>> immutable. For example it now holds onto potentially lazy initialized
>>>>>> fields, such as getEntityKey().
>>>>>> If it's not possible to make it really immutable (all final fields),
>>>>>> we'll need to make it threadsafe and question the name I'm proposing.
>>>>>
>>>>>
>>>>> The specific use-case John is interested in does indeed need to be
>>>>> completely thread-safe and fully concurrent.
>>>>>
>>>>>>
>>>>>> LockMode
>>>>>> From a logical perspective of users, one might think that an entity
>>>>>> being "immutable" doesn't necessarily imply I can't lock on it..
>>>>>> right? I'm not sure how far it could be a valid use case, but these
>>>>>> patches are implying that one can't lock an immutable entity anymore,
>>>>>> as the lock state would be as immutable as anything else on the
>>>>>> EntityEntry.
>>>>>> Are we good with that? Alternatively one might need to think to
>>>>>> separate the lock state handling from the EntityEntry.
>>>>>
>>>>>
>>>>> Conceptually there is nothing wrong with requesting a READ lock on an
>>>>> immutable entity. But like you said, what is the logical expectation there?
>>>>> IMO there should be none. But if we decide that it is OK to req
>>>>
>>>>
>>>> But if we decide that it is OK to request a lock on an immutable entity,
>>>> that is problematic for the idea of a "SharedEntityEntry" or an
>>>> "ImmutableEntityEntry" because a lock is associated with a Session which is
>>>> what the EntityEntry is meant to model... an entity's information in
>>>> relation to a Session. Aka, it now needs to hold state.
>>>>
>>>>
>>>>>
>>>>> Notice I said immutable here and not READ_ONLY which is a specific
>>>>> distinction which is important to other parts of your email that I will
>>>>> address in a second email.
>>>>>
>>>>>
>>>>>>
>>>>>>
>>>>>> On smaller details:
>>>>>> # org.hibernate.engine.internal.SharedEntityEntry is hosted in an
>>>>>> .internal package, I don't think it's right to refactor all the public
>>>>>> API javadoc which was referring to EntityEntry to now refer to the
>>>>>> internal implementation.
>>>>>> # things like EntityEntryExtraState should probably get moved to
>>>>>> .internal packages as well now - we couldn't do that before without
>>>>>> breaking either encapsulation or APIs.
>>>>
>>>>
>>>> +1
>>>>
>>>>>>
>>>>>>
>>>>>> In terms of git patches, the complexity of the changeset risks to get
>>>>>> a bit our of hand. What about we focus on creating a clean pull
>>>>>> request which focuses exclusively on making EntityEntry an interface,
>>>>>> and move things to the right packages and javadoc?
>>>>
>>>>
>>>> Agree 100%
>>>>
>>>>>>
>>>>>> You'd have a trivial EntitEntryFactory, and we can then build the
>>>>>> evolution on top of that, not least maybe helping out by challenging
>>>>>> some points in parallel work.
>>>>>> These are the things I'd leave for a second iteration:
>>>>>> - add various implementations of EntityEntry iteratively, as needed
>>>>>> - the strategy such a Factory would use the pick an implementation
>>>>>> - ultimately, make it possible for an integrator to override such a
>>>>>> Factory
>>>>
>>>>
>>>> This all seems reasonable. For 5.0 I think the most important thing is
>>>> to nail down the idea of EntityEntry as an interface, introduce a Factory
>>>> for building them and agree on the signature for building them. Granted we
>>>> may need iteration to finalize the actual Factory signature, especially as
>>>> OGM finally starts to use it, but I think that in general we can get it
>>>> pretty close. Worst case we just pass high-level constructs like the
>>>> EntityPersister (which OGM supplies custom impls) and the Session and all
>>>> the "EntityEntry state". For the purpose of starting a discussion:
>>>>
>>>> public interface EntityEntryFactory {
>>>> public EntityEntry createEntityEntry(
>>>> PersistenceContext persistenceContext,
>>>> EntityPersister persister,
>>>> Status status,
>>>> Serializable id,
>>>> Object version,
>>>> Object[] loadedState,
>>>> Object rowId,
>>>> LockMode lockMode,
>>>> boolean existsInDatabase,
>>>> boolean disableVersionIncrement,
>>>> boolean lazyPropertiesAreUnfetched);
>>>> }
>>>>
>>>> I would suggest a "creation context" method param object to pass in
>>>> here, but seeing as how we are trying to stop instantiations...
>>>>
>>>> I would prefer the definition of the EntityEntryFactory to use be
>>>> defined via SessionFactoryBuilder interface. I already have plans to have a
>>>> auto-loaded hook for integrators to be able to adjust the
>>>> SessionFactoryBuilder. But as a Service is fine too.
>>>>
>>>>>>
>>>>>> For example with Hibernate OGM we might want to override / re
>>>>>> configure the factories to use custom EntityEntry implementations -
>>>>>> requirements are not fully clear at this point but it seems likely.
>>>>>>
>>>>>> The priority being to define the API as that would be a blocker for
>>>>>> 5.0, we have then better choices to leave more smarter and advanced
>>>>>> EntityEntry implementations for the future; we'd still need to
>>>>>> implement at least the essential ones to make sure the API of the
>>>>>> EntityEntryFactory has all the context it needs.
>>>>>>
>>>>>> Thanks,
>>>>>> Sanne
>>>>>>
>>>>>>
>>>>>> On 24 March 2015 at 09:27, John O'Hara <johara at redhat.com> wrote:
>>>>>> > Steve,
>>>>>> >
>>>>>> > Have you had chance to look at this? Do you have any
>>>>>> > comments/observations?
>>>>>> >
>>>>>> > Thanks
>>>>>> >
>>>>>> > John
>>>>>> >
>>>>>> >
>>>>>> > On 17/03/15 09:24, John O'Hara wrote:
>>>>>> >
>>>>>> > Steve,
>>>>>> >
>>>>>> > I have been having a think about the EntityEntry interface, and have
>>>>>> > forked
>>>>>> > a branch here:
>>>>>> >
>>>>>> >
>>>>>> > https://github.com/johnaoahra80/hibernate-orm/tree/EntityEntryInterface
>>>>>> >
>>>>>> > I know it is nowhere near complete, but was this the sort of idea
>>>>>> > you had in
>>>>>> > mind?
>>>>>> >
>>>>>> > Thanks
>>>>>> >
>>>>>> > John
>>>>>> >
>>>>>> >
>>>>>> > On 13/03/15 09:44, John O'Hara wrote:
>>>>>> >
>>>>>> > EntityEntry retains a reference to a persistenceContext internally
>>>>>> > that
>>>>>> > org.hibernate.engine.spi.EntityEntry#setReadOnly makes calls to, is
>>>>>> > this
>>>>>> > where the session reference is kept? As
>>>>>> > org.hibernate.engine.spi.PersistenceConext is an interface could we
>>>>>> > have a
>>>>>> > different implementation for this use case? e.g. an
>>>>>> > ImmutablePersistenceContext that could be shared across sessions?
>>>>>> >
>>>>>> > For the bytecode enhancement, could we change the enhancer so that
>>>>>> > it adds
>>>>>> > an EntityEntry interface with javassist.
>>>>>> > ClassPool.javassist.ClassPool.makeInterface()() as opposed to adding
>>>>>> > a class
>>>>>> > javassist.ClassPool.makeClass()? I need to have a look at javassit
>>>>>> > to
>>>>>> > confirm what javassist.ClassPool.makeInterface() does.
>>>>>> >
>>>>>> > Thanks
>>>>>> >
>>>>>> > John
>>>>>> >
>>>>>> > On 12/03/15 18:52, Steve Ebersole wrote:
>>>>>> >
>>>>>> > It is possible. Although some of the changes are particularly
>>>>>> > painful.
>>>>>> > Most of EntityEntry, if it is an interface, can be made to work with
>>>>>> > your
>>>>>> > use case. org.hibernate.engine.spi.EntityEntry#setReadOnly I think
>>>>>> > is the
>>>>>> > one exception, because:
>>>>>> > 1) your use case needs it
>>>>>> > 2) it expects the Session to be available internally (its not
>>>>>> > passed)
>>>>>> >
>>>>>> > The bigger thing I am worried about for you is the bytecode stuff,
>>>>>> > as that
>>>>>> > ties very tightly with EntityEntry.
>>>>>> >
>>>>>> _______________________________________________
>>>>>> hibernate-dev mailing list
>>>>>> hibernate-dev at lists.jboss.org
>>>>>> https://lists.jboss.org/mailman/listinfo/hibernate-dev
>>>>>
>>>>>
>>>>
>>>
>>>
>>>
>>> --
>>> John O'Hara
>>> johara at redhat.com
>>>
>>> JBoss, by Red Hat
>>> Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod
>>> Street, Windsor, Berkshire, SI4 1TE, United Kingdom.
>>> Registered in UK and Wales under Company Registration No. 3798903
>>> Directors: Michael Cunningham (USA), Charlie Peters (USA), Matt Parsons
>>> (USA) and Michael O'Neill (Ireland).
>>>
>>
>
>
>
> --
> John O'Hara
> johara at redhat.com
>
> JBoss, by Red Hat
> Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod Street,
> Windsor, Berkshire, SI4 1TE, United Kingdom.
> Registered in UK and Wales under Company Registration No. 3798903 Directors:
> Michael Cunningham (USA), Charlie Peters (USA), Matt Parsons (USA) and
> Michael O'Neill (Ireland).
>
More information about the hibernate-dev
mailing list