[hibernate-issues] [Hibernate-JIRA] Updated: (METAGEN-85) MixedConfigurationTest failing

[ https://hibernate.onjira.com/browse/METAGEN-85?page=com.atlassian.jira.pl... ] Hardy Ferentschik updated METAGEN-85: ------------------------------------- Description: This bug is related to this thread of the mailing list: http://lists.jboss.org/pipermail/hibernate-dev/2012-March/007894.html {quote} While working on Hibernate JPA Modelgen, I noticed that there are test failures in _MixedConfigurationTest_. I thought it was because of my own changes but I also have this problem with a fresh git clone from the official repo: {quote} {noformat} Running org.hibernate.jpamodelgen.test.mixedmode.MixedConfigurationTest Tests run: 4, Failures: 4, Errors: 0, Skipped: 0, Time elapsed: 2.494 sec <<< FAILURE! Results : Failed tests: testAccessTypeForXmlConfiguredEmbeddables(org.hibernate.jpamodelgen.test.mixedmode.MixedConfigurationTest): org.hibernate.jpamodelgen.test.mixedmode.Coordinates_ was not generated. testDefaultAccessTypeApplied(org.hibernate.jpamodelgen.test.mixedmode.MixedConfigurationTest): org.hibernate.jpamodelgen.test.mixedmode.Vehicle_ was not generated. testExplicitXmlConfiguredAccessTypeApplied(org.hibernate.jpamodelgen.test.mixedmode.MixedConfigurationTest): org.hibernate.jpamodelgen.test.mixedmode.Vehicle_ was not generated. testMixedConfiguration(org.hibernate.jpamodelgen.test.mixedmode.MixedConfigurationTest): org.hibernate.jpamodelgen.test.mixedmode.RentalCar_ was not generated. Tests run: 4, Failures: 4, Errors: 0, Skipped: 0 {noformat} _AnnotationMetaEntity.mergeInMembers_: the merged in members aren't really affected to the new entity as the hostingEntity of the attribute is still the original entity. In the case of the _ZeroCoordinates_ entity, it leads to a compilation error as the _SingularAttribute_ import isn't added to the right context (it's added to the original _XmlMetaEmbeddable_ context instead of the _AnnotationEmbeddable_ context this attribute is attached to at the end of the annotation processing). The content of the generated class is as follows: {code} package org.hibernate.jpamodelgen.test.mixedmode; import javax.annotation.Generated; import javax.persistence.metamodel.StaticMetamodel; @Generated(value = "org.hibernate.jpamodelgen.JPAMetaModelEntityProcessor") @StaticMetamodel(ZeroCoordinates.class) public abstract class ZeroCoordinates_ { public static volatile SingularAttribute<ZeroCoordinates, Float> longitude; public static volatile SingularAttribute<ZeroCoordinates, Float> latitude; } {code} I was wondering if providing the way to overwrite the hostingEntity (via removing the final and adding a setter) would be acceptable or not? Even if I do so (draft patch attached), I still have a test failing claiming that _ZeroCoordinates_ shouldn't have any attributes generated. IMHO, there are 2 problems here: * we need to decide if having the attributes generated for _ZeroCoordinates_ is the wanted behavior. Considering that they are explicitly declared in the XML mapping, I think so but, as there is an explicit test checking that they aren't, I'm not sure it's the expected behavior; * if we fix the above problem by removing the attributes from _ZeroCoordinates_, we need to write a test to show the problem of _AnnotationMetaEntity.mergeInMembers_ and fix it (either with the patch provided or with another one doing the same thing). Depending on what we decide, I can probably prepare a pull request for this problem if you want it. Guillaume was: This bug is related to this thread of the mailing list: http://lists.jboss.org/pipermail/hibernate-dev/2012-March/007894.html ******** While working on Hibernate JPA Modelgen, I noticed that there are test failures in MixedConfigurationTest . I thought it was because of my own changes but I also have this problem with a fresh git clone from the official repo: Running org.hibernate.jpamodelgen.test.mixedmode.MixedConfigurationTest Tests run: 4, Failures: 4, Errors: 0, Skipped: 0, Time elapsed: 2.494 sec <<< FAILURE! Results : Failed tests: testAccessTypeForXmlConfiguredEmbeddables(org.hibernate.jpamodelgen.test.mixedmode.MixedConfigurationTest): org.hibernate.jpamodelgen.test.mixedmode.Coordinates_ was not generated. testDefaultAccessTypeApplied(org.hibernate.jpamodelgen.test.mixedmode.MixedConfigurationTest): org.hibernate.jpamodelgen.test.mixedmode.Vehicle_ was not generated. testExplicitXmlConfiguredAccessTypeApplied(org.hibernate.jpamodelgen.test.mixedmode.MixedConfigurationTest): org.hibernate.jpamodelgen.test.mixedmode.Vehicle_ was not generated. testMixedConfiguration(org.hibernate.jpamodelgen.test.mixedmode.MixedConfigurationTest): org.hibernate.jpamodelgen.test.mixedmode.RentalCar_ was not generated. Tests run: 4, Failures: 4, Errors: 0, Skipped: 0 AnnotationMetaEntity.mergeInMembers: the merged in members aren't really affected to the new entity as the hostingEntity of the attribute is still the original entity. In the case of the ZeroCoordinates entity, it leads to a compilation error as the SingularAttribute import isn't added to the right context (it's added to the original XmlMetaEmbeddable context instead of the AnnotationEmbeddable context this attribute is attached to at the end of the annotation processing). The content of the generated class is as follows: package org.hibernate.jpamodelgen.test.mixedmode; import javax.annotation.Generated; import javax.persistence.metamodel.StaticMetamodel; @Generated(value = "org.hibernate.jpamodelgen.JPAMetaModelEntityProcessor") @StaticMetamodel(ZeroCoordinates.class) public abstract class ZeroCoordinates_ { public static volatile SingularAttribute<ZeroCoordinates, Float> longitude; public static volatile SingularAttribute<ZeroCoordinates, Float> latitude; } I was wondering if providing the way to overwrite the hostingEntity (via removing the final and adding a setter) would be acceptable or not? ******** Even if I do so (draft patch attached), I still have a test failing claiming that ZeroCoordinates shouldn't have any attributes generated. IMHO, there are 2 problems here: - we need to decide if having the attributes generated for ZeroCoordinates is the wanted behaviour. Considering that they are explicitely declared in the XML mapping, I think so but, as there is an explicit test checking that they aren't, I'm not sure it's the expected behaviour; - if we fix the above problem by removing the attributes from ZeroCoordinates, we need to write a test to show the problem of AnnotationMetaEntity.mergeInMembers and fix it (either with the patch provided or with another one doing the same thing). Depending on what we decide, I can probably prepare a pull request for this problem if you want it. -- Guillaume -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira