HHH-19203 Unexpected association fetch triggered by Bean Validation #9804
Conversation
| && Hibernate.isPropertyInitialized( traversableObject, traversableProperty.getName() ); | ||
| && persistenceUnitUtil.isLoaded( traversableObject, traversableProperty.getName() ); |
There was a problem hiding this comment.
Hibernate.isPropertyInitialized() is documented to be a synonym for PersistenceUtil.isLoaded(), so this should in principle not have fixed anything.
There was a problem hiding this comment.
If I understand correctly, the issue is around uninitialized collections. So what we're trying to do here is add an additional check equivalent to Hibernate.isInitialized(collection). Apparently there's an undocumented difference between Hibernate.isPropertyInitialized() and PersistenceUtil.isLoaded() for that collections. At minimum we should document that properly/.
There was a problem hiding this comment.
Then we should maybe also change the documentation, because it's not a synonym.
The PersistenceUnitUtil implementation is aware of the SessionFactory and can hence also check laziness of and initialize managed objects that are not enhanced, whereas the Hibernate class can only deal with bytecode enhanced classes.
This is mostly due to the fact that checking if an attribute is actually initialized, we'd have to be able to access the value of the attribute, which we can't without the SessionFactory, because we don't have a PropertyAccess.
There was a problem hiding this comment.
btw, there's a built-in one that relies on the persistence utils:
There was a problem hiding this comment.
Personally, with all bytecode enhancement settings currently deprecated, I don't see any uses for Hibernate.isInitialized anymore. I think at least a documentation update is in order, and would go as far as deprecating the method. I don't think it's possible to reimplement Hibernate.isPropertyInitialized to align PersistenceUnitUtil.isLoaded without significant side effects.
There was a problem hiding this comment.
Hibernate does not require the use of the bytecode enhancer, and most people don't use it.
There was a problem hiding this comment.
Oh sorry I meant Hibernate.isPropertyInitialized not Hibernate.isInitialized. Without bytecode enhancement, the only time it returns unintialized (false) is when the passed in Object entity itself is a HibernateProxy returned by emf.getReferenceById, which is completely covered by Hibernate.isInitialized.
|
Thanks for your pull request! This pull request appears to follow the contribution rules. › This message was automatically generated. |
...src/test/java/org/hibernate/orm/test/annotations/beanvalidation/LazyPropertiesFetchTest.java
Outdated
Show resolved
Hide resolved
...src/test/java/org/hibernate/orm/test/annotations/beanvalidation/LazyPropertiesFetchTest.java
Outdated
Show resolved
Hide resolved
Kamii0909
force-pushed
the
main
branch
2 times, most recently
from
457f37e to
0f463a7
Compare
|
I made adjustments and slight cleaning with the changes from #9806. I noticed formatting of relevant files is kinda off, but I refrained from formatting everything because I want to keep the change set minimum.
I am not confident in Hibernate internal that I could detect all possibly lazily initialized properties. Aside from associations, I think we also have lazy basic group, but iirc that requires bytecode enhancement, which is well covered already? |
|
Hey @marko-bekhta would you mind checking if this makes sense, and checking whether there are any unexpected difference with |
5 participants
More details on Hibernate forum.
By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license
and can be relicensed under the terms of the LGPL v2.1 license in the future at the maintainers' discretion.
For more information on licensing, please check here.
https://hibernate.atlassian.net/browse/HHH-19203