3.1 M2 introduced a regression that causes false positives during
@Configuration class candidate checks. Now performing a call to
AnnotationMetadata#isInterface in addition to checks for @Component and
@Bean annotations when determining whether a candidate is a 'lite'
configuration class. Annotations are in the end interfaces, so both
are filtered out at once.
Issue: SPR-8761
Prior to this change, any instantiation of an
AnnotationConfigApplicationContext would trigger the creation of three
StandardEnvironment objects: one for the ApplicationContext, one for the
AnnotatedBeanDefinitionReader, and one for the
ClassPathBeanDefinitionScanner.
The latter two are immediately swapped out when the ApplicationContext
delegates its environment to these subordinate objects anyway. Not only
is it inefficient to create these two extra Environments, it creates
confusion when debug-level logging is turned on. From the user's
perspective and in practice, there is only one Environment; logging
should reflect that.
This change ensures that only one Environment object is ever created for
a given ApplicationContext. If an AnnotatedBeanDefinitionReader or
ClassPathBeanDefinitionScanner are used in isolation, e.g. against a
plain BeanFactory/BeanDefinitionRegistry, then they will still create
their own local StandardEnvironment object.
All public API compatibility is preserved; new constructors have been
added where necessary to accommodate passing an Environment directly
to ABDR ar CPBDS.
Allowing beans of primitive type to be looked up via getBean(Class), or
to be injected using @Autowired or @Injected or @Resource. Prior to
these changes, an attempt to lookup or inject a bean of, for example,
type boolean would fail because all spring beans are Objects, regardless
of initial type due to the way that ObjectFactory works.
Now these attempts to lookup or inject primitive types work, thanks to
simple changes in AbstractBeanFactory using ClassUtils#isAssignable
methods instead of the built-in Class#isAssignableFrom. The former takes
into account primitives and their object wrapper types, whereas the
latter does not.
The need to declare, look up or inject primitive-typed beans is probably
low -- how often does one need a bean of type boolean or int after all?.
Prior to the introduction of @Bean methods in Spring 3.0, it was not
possible in practice to register primitive beans, so this issue never
came up. Now that one can declare primitive-typed beans, it does make
sense that we properly support by-type lookup and injection without
forcing the user to work with object wrappers.
Issue: SPR-8874
The registration of more than one ConfigurationClassPostProcessor
results in the double-enhancement of @Configuration classes, i.e. a
two-deep CGLIB subclass hierarchy is created.
As a side-effect of changes introduced in 3.1 M2 fixing SPR-8080, this
behavior now results in an infinite loop at CGLIB callback processing
time, leading to a StackOverflowException which is then suppressed by
the container, and ultimately results in the user being presented with
an unintuitive "Bean 'x' is not already in creation" exception.
This fix introduces a marker interface 'EnhancedConfiguration' to be
implemented by all generated @Configuration subclasses. The
configuration class enhancer can then behave in an idempotent fashion
by checking to see whether a candidate @Configuration class is already
assignable to this type i.e. already enhanced and ignore it if so.
Naturally, users should avoid registering more than one
ConfigurationClassPostProcessor, but this is not always possible. As
with the case in point, SPR-8824 originates from problems with
spring-data-neo4j, which explicitly registers its own
ConfigurationClassPostProcessor. The user has little control over this
arrangement, so it is important that the framework is defensive as
described above.
Issue: SPR-8824
Separate concerns of @Configuration class selection from the need to
register certain infrastructure beans such as auto proxy creators.
Prior to this change, ImportSelector implementations were responsible
for both of these concerns, leading to awkwardness and duplication.
Also introduced in this change is ImportBeanDefinitionRegistrar and
two implementations, AutoProxyRegistrar and AspectJAutoProxyRegistrar.
See the refactored implementations of CachingConfigurationSelector,
TransactionManagementConfigurationSelector to see the former;
AspectJAutoProxyConfigurationSelector to see the latter.
ImportSelector and ImportBeanDefinitionRegistrar are both handled as
special-case arguments to the @Import annotation within
ConfigurationClassParser.
These refactorings are important because they ensure that Spring users
will be able to understand and extend existing @Enable* annotations
and their backing ImportSelector and @Configuration classes, as well
as create their own with a minimum of effort.
Also eliminate all 'cache definition' language in favor of
'cache operation' in comments, method and parameter names (most
classes had already been refactored to this effect).
Unfortunately creates a large diff due to whitespace changes as well as
false attribution of authorship from a git/svn 'blame' perspective.
Be sure to perform diffs using `git diff -w` or `svn diff -w` when
reviewing recent changes to these sources to ignore all whitespace.
In favor of existing #setCacheOperationSources(CacheOperationSource...)
Also polish Javadoc throughout, replacing stale references to
CacheDefinitionSource where appropriate as well as other minor changes
Facilitates type-safe programmatic configuration from @Bean methods:
@Bean
public CacheManager cacheManager() {
SimpleCacheManager cm = new SimpleCacheManager();
cm.setCaches(Arrays.asList(
new ConcurrentMapCache("default"),
new ConcurrentMapCache("primary"),
new ConcurrentMapCache("secondary")
));
return cm;
}
Prior to this change, the code above would have raised errors on the
Arrays.asList() call because it returns a Collection<? extends Cache>
as opposed to Collection<Cache>.
After this change, AbstractCacheManager expects
Collection<? extends Cache> throughout.
Refactored getConfig => getApplicationContext such that subclasses have
control over the type of ApplicationContext used by the base class
tests. Done in anticipation of @EnableCaching tests that will favor use
of AnnotationConfigApplicationContext
Also updated all use of ClassPathXmlApplictionContext to
GenericXmlApplicationContext, which is generally preferred.
There was some question about whether enabling subclass proxies via
proxyTargetClass / proxy-target-class settings would break annotation-
based demarcation of joinpoints due to inability to discover those
annotations in various scenarios. The provided tests prove that in
any conceivable case, these annotations (@Transactional, at least)
are discovered in a consistent fashion, meaning that switching proxy
strategies should be transparent to the application and honor
intended annotation semantics.
Anywhere the value of a destroy method may be expressed, specifying
the value "(inferred)" now indicates that the container should attempt
to automatically discover a destroy method. This functionality is
currently limited to detecting public, no-arg methods named 'close';
this is particularly useful for commonly used types such as Hibernate
SessionFactory most JDBC DataSource implementations, JMS connection
factories, and so forth.
This special value is captured as the constant
AbstractBeanDefinition#INFER_METHOD, which in turn serves as the default
value of the @Bean#destroyMethod attribute.
For example in the following case
@Bean
public BasicDataSource dataSource() { ... }
the container will automatically detect BasicDataSource#close and invoke
it when the enclosing ApplicationContext is closed. This is exactly
equivalent to
@Bean(destroyMethod="(inferred)")
public BasicDataSource dataSource() { ... }
A user may override this inference-by-default convention simply by
specifying a different method
@Bean(destroyMethod="myClose")
public MyBasicDataSource dataSource() { ... }
or, in the case of a bean that has an otherwise inferrable 'close'
method, but the user wishes to disable handling it entirely, an empty
string may be specified
@Bean(destroyMethod="")
public MyBasicDataSource dataSource() { ... }
The special destroy method name "(inferred)" may also be specified in
an XML context, e.g.
<bean destroy-method="(inferred)">
or
<beans default-destroy-method="(inferred)">
Note that "(inferred)" is the default value for @Bean#destroyMethod,
but NOT for the destroy-method and default-destroy-method attributes
in the spring-beans XML schema.
The principal reason for introducing this feature is to avoid forcing
@Configuration class users to type destroyMethod="close" every time a
closeable bean is configured. This kind of boilerplate is easily
forgotten, and this simple convention means the right thing is done
by default, while allowing the user full control over customization or
disablement in special cases.
Issue: SPR-8751
Add BridgeMethodResolver#isJava6VisibilityBridgeMethodPair to
distinguish between (a) bridge methods introduced in Java 6 to
compensate for inheriting public methods from non-public superclasses
and (b) bridge methods that have existed since Java 5 to accommodate
return type covariance and generic parameters.
In the former case, annotations should be looked up from the original
bridged method (SPR-7900). In the latter, the annotation should be
looked up against the bridge method itself (SPR-8660).
As noted in the Javadoc for the new method, see
http://stas-blogspot.blogspot.com/2010/03/java-bridge-methods-explained.html
for a useful description of the various types of bridge methods, as
well as http://bugs.sun.com/view_bug.do?bug_id=6342411, the bug fixed in
Java 6 resulting in the introduction of 'visibility bridge methods'.
Issue: SPR-8660, SPR-7900
Remove all convenience variants of #findAllAnnotationAttributes and
refactor the remaining method to accept a MetadataReaderFactory
instead of creating its own SimpleMetadataReaderFactory internally.
This allows clients to use non-default class loaders as well as
customize the particular MetadataReaderFactory to be used (e.g.
'simple' vs 'caching', etc).
Issue: SPR-8752
Removed formal deprecation warnings for AbstractSingletonFactoryBean and
its TransactionProxyFactoryBean and CacheProxyFactoryBean subclasses.
This is principally because TPFB is still used by Grails and could
conceivably be used to good effect by any third-party framework in a
similar fashion. CPFB is new with 3.1, but similar use is predictable.
Deprecations have been replaced by strong recommendations that users
avoid these types in modern Spring applications and favor the use of
namespaces and annotatinos, such as tx: and @Transactional (around
since Spring 2.x) and cache: and @Cacheable.
Issue: SPR-8680, SPR-8686
In anticipation of 'destroy method inference' feature, introduce
ConfigurationClassUtils#INFER_METHOD and update @Bean#destroyMethod to
reflect its use.
Issue: SPR-8751
Cite original inspiriation by Domain-Driven Design, but make clear the
flexible and general-purpose nature of Spring's stereotype annotations
such as @Repository and @Service.
Also update @Repository Javadoc with more explicit instructions about
switching on exception translation through use of
PersistenceExceptionTranslationPostProcessor, and update PETPP Javadoc
for style as well as concrete examples of 'resource factories' that
implement the PersistenceExceptionTranslator interface
Issue: SPR-8691
Prior to this change, StandardServletEnvironment evaluated a
"jndiPropertySourceEnabled" flag to determine whether or not to add a
JndiPropertySource. Following the changes introduced in SPR-8490, there
is now no reason not to enable a JNDI property source by default. This
change eliminates the support for "jndiPropertySourceEnabled" and adds
a JndiPropertySource automatically.
Issue: SPR-8545, SPR-8490
A <scheduled:task> element declared within a
<beans default-lazy-init="true"> element represents a contradiction in
terms: such a task will never be executed.
For this reason, we now override any inherited lazy-init settings
when parsing <scheduled:task> elements, forcing lazy-init to false
for the underlying ScheduledTaskRegistrar bean.
Thanks to Mike Youngstrom for contributing an initial patch.
Issue: SPR-8498
This change returns the invocation order of
ConfigurationClassPostProcessor#enhanceConfigurationClasses to its
pre-3.1 M2 state. An earlier (and now unnecessary) refactoring in
service of @Feature method processing caused the change that this now
reverts.
Prior to this change, an instance of ConfigurationClassPostProcessor
would throw IllegalStateException if its
postProcessBeanDefinitionRegistry method were called more than once.
This check is important to ensure that @Configuration classes are
not proxied by CGLIB multiple times, and works for most normal use
cases.
However, if the same CCPP instance is used to process multiple
registries/factories/contexts, this check creates a false negative
because it does not distinguish between invocations of
postProcessBeanDefinitionRegistry across different registries.
A use case for this, though admittedly uncommon, would be creating
a CCPP instance and registering it via
ConfigurableApplicationContext#addBeanDefinitionPostProcessor against
several ApplicationContexts. In such a case, the same CCPP instance
will post-process multiple different registry instances, and throw the
above mentioned exception.
With this change, CCPP now performs lightweight tracking of the
registries/beanFactories that it has already processed by recording
the identity hashcodes of these objects. This is only slightly more
complex than the previous boolean-based 'already processed' flags, and
prevents this issue (however rare it may be) from occurring.
Issue: SPR-8527
Prior to this change, @EnableTransactionManagement (via the
ProxyTransactionManagementConfiguration class) did not properly
register its auto-proxy creator through the usual AopConfigUtils
methods. It was trying to register the APC as a normal @Bean method,
but this causes issues (SPR-8494) with the logic in
AopConfigUtils#registerOrEscalateApcAsRequired, which expects the APC
bean definition to have a beanClassName property. When the APC is
registered via a @Bean definition, it is actually a
factoryBean/factoryMethod situation with no directly resolvable
beanClass/beanClassName.
To solve this problem, ImportSelector#selectImports has been refactored
to accept an ImportSelector.Context instance. This object contains the
AnnotationMetadata of the importing class as well as the enclosing
BeanDefinitionRegistry to allow for the kind of conditional bean
registration necessary here. In this case, the bean definition that
must be registered conditionally is that of the auto-proxy creator.
It should only be registered if AdviceMode == PROXY, and thus the
ImportSelector is an appropriate place to make this happen. It must
happen as a BeanDefinition (rather than a @Bean method) for
compatibility with AopConfigUtils, and working with the
BeanDefinitionRegistry API allows for that. This change does mean that
in certain cases like this one, #selectImports has container modifying
side effects. Documentation has been updated to reflect.
Issue: SPR-8411, SPR-8494
For the particular use case detailed in SPR-8514, with this change we
now attempt to determine the object type of a FactoryBean through its
generic type parameter if possible.
For (a contrived) example:
@Configuration
public MyConfig {
@Bean
public FactoryBean<String> fb() {
return new StringFactoryBean("foo");
}
}
The implementation will now look at the <String> generic parameter
instead of attempting to instantiate the FactoryBean in order to call
its #getObjectType() method.
This is important in order to avoid the autowiring lifecycle issues
detailed in SPR-8514. For example, prior to this change, the following
code would fail:
@Configuration
public MyConfig {
@Autowired Foo foo;
@Bean
public FactoryBean<String> fb() {
Assert.notNull(foo);
return new StringFactoryBean("foo");
}
}
The reason for this failure is that in order to perform autowiring,
the container must first determine the object type of all configured
FactoryBeans. Clearly a chicken-and-egg issue, now fixed by this
change.
And lest this be thought of as an obscure bug, keep in mind the use case
of our own JPA support: in order to configure and return a
LocalContainerEntityManagerFactoryBean from a @Bean method, one will
need access to a DataSource, etc -- resources that are likely to
be @Autowired across @Configuration classes for modularity purposes.
Note that while the examples above feature methods with return
types dealing directly with the FactoryBean interface, of course
the implementation deals with subclasses/subinterfaces of FactoryBean
equally as well. See ConfigurationWithFactoryBeanAndAutowiringTests
for complete examples.
There is at least a slight risk here, in that the signature of a
FactoryBean-returing @Bean method may advertise a generic type for the
FactoryBean less specific than the actual object returned (or than
advertised by #getObjectType for that matter). This could mean that an
autowiring target may be missed, that we end up with a kind of
autowiring 'false negative' where FactoryBeans are concerned. This is
probably a less common scenario than the need to work with an autowired
field within a FactoryBean-returning @Bean method, and also has a clear
workaround of making the generic return type more specific.
Issue: SPR-8514
isCglibProxy* methods in AopUtils are useful in lower-level modules,
i.e. those that cannot depend on .aop. Therefore copied these methods
to ClassUtils; deprecated the existing ones in AopUtils and now
delegating to the new location; switched all usage of
AopUtils#isCglibProxy* within the framework to use
ClassUtils#isCglibProxy* instead.
Even after applying @Ignore to these tests at the class level, they
still run (and fail) under ant when the jmxremote_optional jar is not
present. See the issues mentioned below for information on how these
tests will be re-enabled.
Issue: SPR-8089, SPR-8093, SPR-8458
Chris Beams
Costin Leau
Juergen Hoeller
Sam Brannen
Rossen Stoyanchev
Keith Donald