HHH-4962: Taking @OneToMany(mappedBy) Into Account

[systemtrap]

@AuditMappedBy can already be used to define an inverse mapping for envers.
Being unable to see any reason why standard annotations may not be taken
into account mappedBy from @OneToMany is used for an inverse @manytoone
mapping and can be overridden by @AuditMappedBy.

[lukasz-antoniak]

Hi Matthias!

Thanks for submitting pull request. Don't you think that it would be more consistent if CollectionMetadataGenerator.searchMappedBy(PersistentClass, Table) method was modified as I did here: lukasz-antoniak@12464d7? Anyway, your changes to AuditedPropertiesReader and/or my to CollectionMetadataGenerator, does not make ImplicitMappedByTest test pass. IMO the type of relation presented in this JIRA issue shall be mapped as "fakeOneToManyBidirectional", and CollectionMetadataGenerator.addCollection() method needs to be adjusted. Could you investigate further?

Regards,
Lukasz Antoniak

[lukasz-antoniak]

I think that I have solved this issue. Please see another commit: lukasz-antoniak@19c7a81.

The @OneToMany.mappedBy() attribute has to be read as Matthias suggests. Otherwise it is difficult to differ bi-directional relation with @jointable on the owning @manytoone side from other relations (for example OneToManyComponentTestEntity, ParentReferencingChildTest, BasicList, IdMapKey, BasicMap, BasicWhereJoinTable, ManyToManyInverseToSuperclassTest etc. tests fail) inside CollectionMetadataGenerator.addCollection() method.

Adam, if you have time please check my findings.

Regards,
Lukasz Antoniak

[adamw]

Ah, this is indeed a case not covered yet :)

And it must be in fact treated same as a "fake" relation. Code looks good, if the new test passes, and the old test as well, push it ;)

[lukasz-antoniak]

Issue resolved. Thanks Matthias.