[hibernate-issues] [Hibernate-JIRA] Issue Comment Edited: (HHH-7158) Regression: null values on NaturalId's at persist-time are causing subsequent NPE's and wrong behaviour

Regression: null values on NaturalId's at persist-time are causing subsequent NPE's and wrong behaviour ------------------------------------------------------------------------------------------------------- Key: HHH-7158 URL: https://hibernate.onjira.com/browse/HHH-7158 Project: Hibernate ORM Issue Type: Bug Components: core Affects Versions: 4.1.1 Environment: Hibernate 4.1.1, db-indipendent (hsqldb used in testcase) Reporter: Guenther Demetz Labels: naturalId Attachments: TestCasePersistNullNaturalId.jar Original Estimate: 2h Remaining Estimate: 2h Since Hibernate 4.1.1 it is no more possible to have uninitialized NaturalId values when persisting, in older versions this was always possible (it was even possible to search NaturalId by null-values). The problem stays in the new method NaturalIdXrefDelegate$NaturalIdResolutionCache#areSame(...) where org.hibernate.type.Type#compare(...) is used instead of org.hibernate.type.Type#isEqual(...) Proposed solution (method NaturalIdResolutionCache#areSame): if ( !naturalIdTypes[i].isEqual(naturalIdValues[i], values[i], factory)) { return false; } instead of: if ( naturalIdTypes[i].compare( naturalIdValues[i], values[i] ) != 0 ) { return false; } This makes the first test-case green (see attached Testcase). The second test-case becomes green, due making method EntityType#isEqual(Object x, Object y, SessionFactoryImplementor factory) stable against null x,y values. That sould be trivial. N.B.: Please consider also that the EntityType#compare(Object x, Object y) implementation currently always returns 0 (there's a TODO-note), whilst EntityType#isEqual(Object x, Object y, SessionFactoryImplementor factory) seems to be a good implementation. Last but not least: isEqual methods should also be a little bit faster than compare-method, cause the latter must calculate differences.