Prior to this commit, the BeanOverrideBeanFactoryPostProcessor rejected
any attempt to override a non-singleton bean; however, due to interest
from the community, we have decided to provide support for overriding
non-singleton beans via the Bean Override mechanism — for example, when
using @MockitoBean, @MockitoSpyBean, and @TestBean.
With this commit, we now support Bean Overrides for non-singletons: for
standard JVM runtimes as well as AOT processing and AOT runtimes. This
commit also documents that non-singletons will effectively be converted
to singletons when overridden and logs a warning similar to the
following.
WARN: BeanOverrideBeanFactoryPostProcessor - Converting 'prototype' scoped bean definition 'myBean' to a singleton.
See gh-33602
See gh-32933
See gh-33800
Closes gh-35574
At some point, TestNG started implementing listener methods as interface
default methods, so we no longer need to override those methods with
empty implementations.
Prior to this commit, AbstractTestNGSpringContextTests was not
thread-safe with regard to tracked exceptions.
To address that, AbstractTestNGSpringContextTests now tracks the test
exception via a ThreadLocal.
Closes gh-35528
Since Spring Framework 4.2, DefaultContextCache supported an LRU (least
recently used) eviction policy via a custom LruCache which extended
LinkedHashMap. The LruCache reacted to LinkedHashMap's
removeEldestEntry() callback to remove the LRU context if the maxSize
of the cache was exceeded.
Due to the nature of the implementation in LinkedHashMap, the
removeEldestEntry() callback is invoked after a new entry has been
stored to the map.
Consequently, a Spring ApplicationContext (C1) was evicted from the
cache after a new context (C2) was loaded and added to the cache,
leading to failure scenarios such as the following.
- C1 and C2 share an external resource -- for example, a database.
- C2 initializes the external resource with test data when C2 is loaded.
- C1 cleans up the external resource when C1 is closed.
- C1 is loaded and added to the cache.
- C2 is loaded and added to the cache before C1 is evicted.
- C1 is evicted and closed.
- C2 tests fail, because C1 removed test data required for C2.
To address such scenarios, this commit replaces the custom LruCache
with custom LRU eviction logic in DefaultContextCache and revises
the put(MergedContextConfiguration, ApplicationContext) method to
delegate to a new evictLruContextIfNecessary() method.
This commit also introduces a new put(MergedContextConfiguration,
LoadFunction) method in the ContextCache API which is overridden by
DefaultContextCache to ensure that an evicted context is removed and
closed before a new context is loaded to take its place in the cache.
In addition, DefaultCacheAwareContextLoaderDelegate has been revised to
make use of the new put(MergedContextConfiguration, LoadFunction) API.
Closes gh-21007
This commit updates the `HttpStatus` enum with the latest changes in
RFC9110:
* deprecate "413 Payload Too Large" in favor of "413 Content Too Large"
* deprecate "418 I'm a teapot" as it was meant as a joke and is now
marked as unused
* Introduce new "421 Misdirected Request"
* deprecate "422 Unprocessable Entity" in favor of
"422 Unprocessable Content"
* deprecate "509 Bandwidth Limit Exceeded" as it's now unassigned
* deprecate "510 Not Extended" as it's now marked as "historic"
The relevant exceptions, test matchers and more have been updated as a
result.
Closes gh-32870
This commit updates Jackson 3 JSON support to use JsonMapper
instead of ObjectMapper in converters, codecs and view constructors.
As a consequence, AbstractJacksonDecoder, AbstractJacksonEncoder,
AbstractJacksonHttpMessageConverter and JacksonCodecSupport are
now parameterized with <T extends ObjectMapper>.
Closes gh-35282
This commit reorders and clarifies the usage instructions for
ApplicationEvents to:
1. Recommend method parameter injection as the primary approach, since
ApplicationEvents has a per-method lifecycle
2. Clarify that ApplicationEvents is not a general Spring bean and
cannot be constructor-injected
3. Explicitly state that field injection is an alternative approach
This addresses confusion where developers expect ApplicationEvents to
behave like a regular Spring bean eligible for constructor injection.
See gh-35297
Closes gh-35335
Signed-off-by: khj68 <junthewise@gmail.com>
Sam Brannen
Stéphane Nicoll
rstoyanchev
Réda Housni Alaoui
Manu Sridharan
Sébastien Deleuze
Johnny Lim
Brian Clozel
khj68
Juergen Hoeller
NeatGuyCoding
Rob Worsnop