Previously, if a factory method is defined on a parent, the generated
code would blindly use the method's declaring class for both the target
of the generated code, and the signature of the method.
This commit improves the resolution by considering the factory metadata
in the BeanDefinition.
Closes gh-32609
This commit improves on the bean overriding feature in several ways:
the API is simplified and polished (metadata and processor contracts,
etc...).
The commit also reworks infrastructure classes (context customizer,
test execution listener, BeanOverrideBeanFactoryPostProcessor, etc...).
Parsing of annotations is now fully stateless.
In order to avoid OverrideMetadata in bean definition and to make a
first step towards AOT support, the BeanOverrideBeanFactoryPostProcessor
now delegates to a BeanOverrideRegistrar to track classes to parse,
the metadata-related state as well as for the field injection methods
for tests.
Lastly, this commit increases the test coverage for the provided
annotations and adds integration tests and fixes a few `@TestBean`
issues.
This commit makes the use of bean definition overriding more visible and
prepare for a deprecation of the feature in the next major release.
As of this commit, use of bean definition overriding logs at INFO level.
The previous log level can be restored by setting the
allowBeanDefinitionOverriding flag explicitly on the BeanFactory (or
via the related ApplicationContext).
A number of tests that are using bean overriding on purpose have been
updated to set this flag, which will make them easier to find once we
actually deprecate the feature.
Closes gh-31288
The ApplicationContext has a resource cache that it uses while the
context is being refreshed. Once the refresh phase completes, the
cache is cleared as its content is no longer in use. If beans are
lazily initialized, there is a case where the cache might be filled
again as processing is deferred past the refresh phase.
For those cases, the cache is not cleared. This can be a problem when
running in this mode with a large set of test configurations as the TCF
caches the related contexts.
This commit includes an additional TestExecutionListener to the TCF that
clears the resource cache if necessary once a test class completes. It
is ordered so that it runs after `@DirtiesContext` support as marking
the context as dirty will remove it from the cache and make that extra
step unnecessary.
Closes gh-30954
This commit adds a new way to use MockMvc by returning a MvcResult that
is AssertJ compatible. Compared to the existing MockMvc infrastructure,
this new model has the following advantages:
* Can be created from a MockMvc instance
* Dedicated builder methods for easier setup
* Assertions use familiar AssertJ syntax with discoverable assertions
through a DSL
* Improved exception message
See gh-21178
Co-authored-by: Brian Clozel <brian.clozel@broadcom.com>
This commit adds AssertJ compatible assertions for HTTP request and
responses, including headers, media type, and response body. The latter
delegate to the recently included JSON assert support.
See gh-21178
Historically, we have rarely intentionally thrown a
NullPointerException in the Spring Framework. Instead, we prefer to
throw either an IllegalArgumentException or IllegalStateException
instead of a NullPointerException.
However, changes to the code in recent times have introduced the use of
Objects.requireNonNull(Object) which throws a NullPointerException
without an explicit error message.
The latter ends up providing less context than a NullPointerException
thrown by the JVM (since Java 14) due to actually de-referencing a
null-pointer. See https://openjdk.org/jeps/358.
In light of that, this commit revises our current use of
Objects.requireNonNull(Object) by removing it or replacing it with
Assert.notNull().
However, we still use Objects.requireNonNull(T, String) in a few places
where we are required to throw a NullPointerException in order to
comply with a third-party contract such as Reactive Streams.
Closes gh-32430
This commit upgrades to a major new release of HtmlUnit. This is a
breaking change as HtmlUnit 3 moves to a `org.htmlunit` package and
calling code needs to be restructured.
Our use of Selenium has been adapted accordingly, moving to Selenium
3, using the new org.seleniumhq.selenium:htmlunit3-driver integration.
Closes gh-30392
Sam Brannen
Sébastien Deleuze
Stéphane Nicoll
Arjen Poutsma
Arjen Poutsma
Simon Baslé
Juergen Hoeller
rstoyanchev
PhilKes
Anton Ždanov