JpaMetamodelMappingContext的目的是什么？

Question

spring-data-jpa creates an instance of JpaMetamodelMappingContext , which maintains a metamodel of JPA entities, when @EnableJpaRepositories is used.

We'd like to use the CrudRepository features ( findOne , delete ...) in conjunction with org.springframework.data.jpa.domain.Specifications<T> . @Query or the dynamically generated queries from method names are not needed.

Do we need the JpaMetamodelMappingContext in this scenario?

(Reason: that context is rather huge in terms of heap space. If it's not needed for our use cases, we'd try to skip its creation.)

Thanks a lot!

Answer 1

Just for your reference, after one your running in live environment: It's possible to run spring-data's CrudRepository and JpaRepository without the JPA metamodel. Our Configuration looks like:

@Configuration
@EnableJpaRepositories(basePackageClasses = Foo.class)
@ComponentScan(basePackageClasses = { Bar.class })
public class DomainConfiguration {

    /**
     * The returned factory will built a {@link JpaMetamodelMappingContext} without a JPA MetaModel.
     *
     * This could save a couple of hundres MBs of heap space, but may disable some of spring-data's features.
     * 
     * @return factory bean for {@link JpaMetamodelMappingContext}.
     */
    @Bean(name = BeanDefinitionNames.JPA_MAPPING_CONTEXT_BEAN_NAME)
    public JpaMetamodelMappingContextFactoryBean jpaMetamodelMappingContextFactoryBean() {
        JpaMetamodelMappingContextFactoryBean factory = new JpaMetamodelMappingContextFactoryBean();
        return factory;
    }
}