This is meant as a guide for writing test cases to be attached to bug reports in the Hibernate Jira. Really most of the information here works just as well when asking for help on community help channels (forums, IRC, HipChat, etc).
There are a number of tenants that make up a good test case as opposed to a poor one. In fact there are a few guides for this across the web including (MCVE) and (SSCCE). These guides all assert the same ideas albeit using different terms. Given the ubiquity of StackOverflow and the fact that the MCVE guidelines were written specifically for StackOverflow, we will use those terms here as we assume most developers have seen them before:
-
(M)inimal - Provide just the minimal information needed. If second level caching is irrelevant to the bug report then the test should not use second level caching. If entity inheritance is irrelevant then do not use it in the test. If your application uses Spring Data, remove Spring Data from the test.
-
(C)omplete - Provide all information needed to reproduce the problem. If a bug only occurs when using bytecode enhancement, then the test should include bytecode enhancement. In other words the test should be self-contained.
-
(V)erifiable - The test should actually reproduce the problem being reported.
JUnit 5 offers better support for integration, compared to JUnit 4, via extensions. Hibernate builds on those concepts in its hibernate-testing
module allowing set up of test fixtures using annotations. The following sections describe the Hibernate extensions.
Note
|
The extensions exist in the org.hibernate.testing.orm.junit package, as opposed to the older org.hibernate.testing.junit4 package used with JUnit 4.
|
Manages a ServiceRegistry
as part of the test lifecycle. 2 in fact, depending on the annotation(s) used.
@BootstrapServiceRegistry
-
configures Hibernate’s bootstrap
BootstrapServiceRegistry
which manages class-loading, etc.@BootstrapServiceRegistry
is used to provide Java services and HibernateIntegrator
implementations for the test. @ServiceRegistry
-
configures Hibernate’s standard
StandardServiceRegistry
.@ServiceRegistry
is used to provide settings, contributors, services, etc.
Also exposes ServiceRegistryScope
via JUnit 5 ParameterResolver
. ServiceRegistryScope
allows
access to the managed ServiceRegistry
from tests and callbacks.
@BootstrapServiceRegistry(
javaServices=@JavaServices(
role=TypeContributions.class,
impls=CustomTypeContributions.class
),
...
)
@ServiceRegistry(
settings=@Setting(
name="hibernate.show_sql",
value="true"
),
services=@Service(
role=ConnectionProvider.class,
impl=CustomConnectionProvider.class
),
...
)
class TheTest {
@Test void testIt(ServiceRegistryScope scope) {
StandardServiceRegistry reg = scope.getRegistry();
...
}
}
Manages the domain model for the test as part of its lifecycle.
@DomainModel
-
defines the sources of the domain model used in the test - type contributions, managed classes, XML mappings, etc.
If available, this extension uses the ServiceRegistry
instances available from ServiceRegistryExtension.
Exposes DomainModelScope
via JUnit5 ParameterResolver
, allowing access to details about the domain model from the org.hibernate.mapping
"boot model".
@DomainModel(
standardDomainModels=StandardDomainModel.ANIMAL,
annotatedClasses={Entity1.class, Entity2.class},
xmlMappings="resource/path/to/my-mapping.xml",
...
)
class TheTest {
@Test void testIt(DomainModelScope scope) {
MetadataImplementor meta = scope.getDomainModel();
...
PersistentClass entityMapping = scope.getEntityBinding(Entity1.class);
...
scope.withHierarchy(Entity1.class, (entityMapping) -> {
...
}
}
}
Manages a Hibernate SessionFactory
as part of the test lifecycle.
@SessionFactory
-
is used to configure the runtime aspects of the
SessionFactory
fixture.
If available, uses the ServiceRegistry
instances available from ServiceRegistryExtension as well
as the domain model defined by DomainModelExtension.
Exposes SessionFactoryScope
via JUnit5 ParameterResolver
.
@SessionFactory(
generateStatistics=true,
exportSchema=true,
useCollectingStatementInspector=true,
...
)
class TheTest {
@Test void testIt(SessionFactoryScope scope) {
SQLStatementInspector sqlCollector = scope.getCollectingStatementInspector();
sqlCollector.clear();
scope.inTransaction( (session) -> {
...
assertThat(sqlCollector.getSqlQueries()).isEmpty();
} );
Entity1 e = scope.fromTransaction( (session) -> {
Entity1 it = session.find(Entity1.class, id);
...
return it;
} );
}
}
Allows filtering tests based on Dialect used. Implemented as a JUnit ExecutionCondition
which is used to dynamically determine whether a test should be run. Used in conjunction with:
@RequiresDialect
-
says to only run this test for the given Dialect(s).
@SkipForDialect
-
says to skip this test for the given Dialect(s).
Used with @ExpectedException
to allow testing that an excepted exception occurs as the "success" condition.
@DomainModel(...)
@SessionFactory(...)
class TheTest {
@Test
@ExpectedException(UnknownEntityTypeException.class)
void testIt(SessionFactoryScope) {
scope.inTransaction( (session) -> {
// Should fail as MyEmbeddable is not an entity
session.find(MyEmbeddable.class, 1);
} );
}
}
Used with @FailureExpected
to indicate that a test is (currently) expected to fail. You might use this, e.g., for a test that is the reproducer for a bug report before working on it. It basically just flips the success/failure condition. In fact, a test marked with @FailureExpected
will be marked a failure if it succeeds.
@Test
@JiraKey("HHH-123456789")
@FailureExpected
void bugReproducer(...) {...}
Both are used for testing log messages.
@LoggingInspections
-
used to watch more than one "message key".
MessageKeyInspection
-
used to watch a single "message key".
Used in conjunction with @Jpa
to build tests with an EntityManagerFactory
fixture.
Since Hibernate’s SessionFactory
is a EntityManagerFactory
, @BootstrapServiceRegistry
, @ServiceRegistry
, @DomainModel
and @SessionFactory
can also be used to perform tests with a (SessionFactory
as a) EntityManagerFactory
fixture.
The distinction with @Jpa
is that EntityManagerExtension
uses the JPA-defined bootstrap APIs. How the
SessionFactory
is built is the difference.
Historically, Hibernate used JUnit 4 for its test suite. Since the release of JUnit 5, we’ve moved to using the testing approach outlined in JUnit 5 extensions. However, many existing tests still use the legacy JUnit 4 based infrastructure (boilerplate) based on "test templates".
The Hibernate team maintains a set of "test templates" intended to help developers write tests. These test templates are maintained in GitHub @ hibernate-test-case-templates
-
If you want to use the Hibernate native API, you should follow the instructions from this article.
-
If you want to use JPA, you should use the JPA templates that were detailed in this article.
Note
|
the test templates are generally not a good starting point for problems building the SessionFactory/EntityManager. In JUnit terms they manage the SessionFactory/EntityManager as set-up and teardown constructs._ |
When using "test templates" you can annotate a single test or a whole test class with one of the following annotations:
-
FailureExpected - allows to skip a single test or all tests of a class, because test failures are expected. The test will actually run, but not lead to an error report. In fact if a test is marked with
@FailureExpected
and the test actually succeeds, an error occurs. As a parameter to this annotation a jira key is required. -
NotImplementedYet - test classes or methods annotated with @NotImplementedYet will run but not fail if the feature(s) that are being tested are not implemented yet for the current version. Optionally, a message and a version that is expected to have the feature already implemented can be provided as parameters.
-
RequiresDialect - tests methods/classes annotated with
@RequiresDialect
will only run if the current Dialect matches the one specified as annotation parameter. You can also specify a comment and/or jira key explaining why this test requires a certain dialect -
RequiresDialectFeature - tests methods/classes annotated with
@RequiresDialectFeature
will only run if the current Dialect offers the specified feature. Examples for this features areSupportsSequences
,SupportsExpectedLobUsagePattern
orSupportsIdentityColumns
. You can add more features if you need to. Have a look atDialectChecks
. -
SkipForDialect - tests methods/classes annotated with
@SkipForDialect
will not run if the current Dialect matches the one specified as annotation parameter. You can also specify a comment and/or jira key explaining why this test has to be skipped for the Dialect.