Prior to this commit, the codebase was using a mix of log4j.xml
and log4j.properties for test-related logging configuration. This
can be an issue as log4j takes the xml variant first when looking
for a default bootstrap configuration.
In practice, some modules declaring the properties variant were
taking the xml variant configuration from another module.
The general structure of the configuration has also been
harmonized to provide a standard console output as well as an
easy way to enable trace logs for the current module.
Restored original singleton-only adaptInnerBeanName behavior, avoiding endless unique name calculation for every single prototype instance. Non-named inner BeanDefinition objects now suffixed with an identity hex code analogous to non-named XML bean definitions, largely avoiding naming collisions to begin with. After SPR-11246, post-processors can deal with unstable classes per bean name, so occasional collisions aren't a hard problem anymore.
Issue: SPR-11545
This commit continues the work for fixing memory leaks resulting from
CGLIB subclass generation for beans relying on method injection.
- Set proxy callbacks on the CGLIB Factory (i.e., the instance) instead
of in the generated subclass (i.e., via the Enhancer).
- Convert private inner classes in CglibSubclassingInstantiationStrategy
to private static classes in order to avoid unnecessary coupling to
classes generated using CGLIB.
- Tidy up XmlBeanFactoryTests.
- Update logic in serializableMethodReplacerAndSuperclass() so that it
finally aligns with the decision made for SPR-356.
Issue: SPR-10785, SPR-356
Prior to this commit, the inclusion of the 'overloaded' flag in the
implementations of equals() and hashCode() in MethodOverride could lead
to adverse effects in the outcome of equals() in AbstractBeanDefinition.
For example, given two bean definitions A and B that represent the
exact same bean definition metadata for a bean that relies on method
injection, if A has been validated and B has not, then A.equals(B) will
potentially return false, which is not acceptable behavior.
This commit addresses this issue by removing the 'overloaded' flag from
the implementations of equals() and hashCode() for MethodOverride.
Issue: SPR-11420
According to the JDK's documentation and changelog, the Introspector itself safely handles weak references as of JDK 6 update 21 (which is what we require for Spring 4.0).
Issue: SPR-11356
This in particular allows for specifying "spring.getenv.ignore" and "spring.beaninfo.ignore" in a local way within the application, in case that JVM-level system properties are locked.
Issue: SPR-9014
Issue: SPR-11297
This allows us to avoid early initialization of footprint-heavy ConcurrentHashMaps, and reuses the existing postProcessingLock which will typically be active already when registerExternallyManaged* calls come in from the bean factory's post-processing phase.
Issue: SPR-11343
Specifically, read and write methods are allowed to express property types with superclass/subclass relationships in both directions now.
Issue: SPR-11139
Specifically, if the current factory has Qualifier/ContextAnnotationAutowireCandidateResolver set up, it is important to pass it on to ancestor factories to get consistent qualifier matching results.
Issue: SPR-10966
Our per-bean caching in AutowiredAnnotationBeanPostProcessor and co relies on unique bean names, so this change fixes potential cache mismatch problems occuring there.
Issue: SPR-11131
Attempt to improve performance by caching TypeDescriptors against bean
PropertyDescriptors in CachedIntrospectionResults.
This change is an attempt to fix the failing performance test case
`testPrototypeCreationWithOverriddenResourcePropertiesIsFastEnough` in
`AnnotationProcessorPerformanceTests`.
Stephane Nicoll
Juergen Hoeller
Sam Brannen
Phillip Webb