Although the reference documentation listed the new @MVC support
classes and their benefits, it did not explicitly mention a few
use cases that are no longer supported. There is now a specific
section on the new support classes listing exactly what is not
supported.
Similary the @RequestMapping annotation never refered explicitly
to the existence of old and new support and never made it clear
exactly what the differences are.
Both have not been corrected.
SPR-9063, SPR-9042
Uses of AnnotationMetadata#getAnnotationAttributes throughout the
framework have been updated to use the new AnnotationAttributes API in
order to take advantage of the more concise, expressive and type-safe
methods there.
All changes are binary compatible to the 3.1.0 public API, save
the exception below.
A minor binary compatibility issue has been introduced in
AbstractCachingConfiguration, AbstractAsyncConfiguration and
AbstractTransactionManagementConfiguration when updating their
protected Map<String, Object> fields representing annotation attributes
to use the new AnnotationAttributes API. This is a negligible breakage,
however, as the likelilhood of users subclassing these types is very
low, the classes have only been in existence for a short time (further
reducing the likelihood), and it is a source-compatible change given
that AnnotationAttributes is assignable to Map<String, Object>.
Background
Spring 3.1 introduced the @ComponentScan annotation, which can accept
an optional array of include and/or exclude @Filter annotations, e.g.
@ComponentScan(
basePackages = "com.acme.app",
includeFilters = { @Filter(MyStereotype.class), ... }
)
@Configuration
public class AppConfig { ... }
@ComponentScan and other annotations related to @Configuration class
processing such as @Import, @ImportResource and the @Enable*
annotations are parsed using reflection in certain code paths, e.g.
when registered directly against AnnotationConfigApplicationContext,
and via ASM in other code paths, e.g. when a @Configuration class is
discovered via an XML bean definition or when included via the
@Import annotation.
The ASM-based approach is designed to avoid premature classloading of
user types and is instrumental in providing tooling support (STS, etc).
Prior to this commit, the ASM-based routines for reading annotation
attributes were unable to recurse into nested annotations, such as in
the @Filter example above. Prior to Spring 3.1 this was not a problem,
because prior to @ComponentScan, there were no cases of nested
annotations in the framework.
This limitation manifested itself in cases where users encounter
the ASM-based annotation parsing code paths AND declare
@ComponentScan annotations with explicit nested @Filter annotations.
In these cases, the 'includeFilters' and 'excludeFilters' attributes
are simply empty where they should be populated, causing the framework
to ignore the filter directives and provide incorrect results from
component scanning.
The purpose of this change then, is to introduce the capability on the
ASM side to recurse into nested annotations and annotation arrays. The
challenge in doing so is that the nested annotations themselves cannot
be realized as annotation instances, so must be represented as a
nested Map (or, as described below, the new AnnotationAttributes type).
Furthermore, the reflection-based annotation parsing must also be
updated to treat nested annotations in a similar fashion; even though
the reflection-based approach has no problem accessing nested
annotations (it just works out of the box), for substitutability
against the AnnotationMetadata SPI, both ASM- and reflection-based
implementations should return the same results in any case. Therefore,
the reflection-based StandardAnnotationMetadata has also been updated
with an optional 'nestedAnnotationsAsMap' constructor argument that is
false by default to preserve compatibility in the rare case that
StandardAnnotationMetadata is being used outside the core framework.
Within the framework, all uses of StandardAnnotationMetadata have been
updated to set this new flag to true, meaning that nested annotation
results will be consistent regardless the parsing approach used.
Spr9031Tests corners this bug and demonstrates that nested @Filter
annotations can be parsed and read in both the ASM- and
reflection-based paths.
Major changes
- AnnotationAttributes has been introduced as a concrete
LinkedHashMap<String, Object> to be used anywhere annotation
attributes are accessed, providing error reporting on attribute
lookup and convenient type-safe access to common annotation types
such as String, String[], boolean, int, and nested annotation and
annotation arrays, with the latter two also returned as
AnnotationAttributes instances.
- AnnotationUtils#getAnnotationAttributes methods now return
AnnotationAttributes instances, even though for binary compatibility
the signatures of these methods have been preserved as returning
Map<String, Object>.
- AnnotationAttributes#forMap provides a convenient mechanism for
adapting any Map<String, Object> into an AnnotationAttributes
instance. In the case that the Map is already actually of
type AnnotationAttributes, it is simply casted and returned.
Otherwise, the map is supplied to the AnnotationAttributes(Map)
constructor and wrapped in common collections style.
- The protected MetadataUtils#attributesFor(Metadata, Class) provides
further convenience in the many locations throughout the
.context.annotation packagage that depend on annotation attribute
introspection.
- ASM-based core.type.classreading package reworked
Specifically, AnnotationAttributesReadingVisitor has been enhanced to
support recursive reading of annotations and annotation arrays, for
example in @ComponentScan's nested array of @Filter annotations,
ensuring that nested AnnotationAttributes objects are populated as
described above.
AnnotationAttributesReadingVisitor has also been refactored for
clarity, being broken up into several additional ASM
AnnotationVisitor implementations. Given that all types are
package-private here, these changes represent no risk to binary
compatibility.
- Reflection-based StandardAnnotationMetadata updated
As described above, the 'nestedAnnotationsAsMap' constructor argument
has been added, and all framework-internal uses of this class have
been updated to set this flag to true.
Issue: SPR-7979, SPR-8719, SPR-9031
Ensure that both FlashMapManager methods - the one invoked at the
start of a request and the one invoked before a redirect, update
the underlying storage fully since it's not guaranteed that both
will be invoked on any given request.
Also move the logic to remove expired FlashMap instances to the
metohd invoked at the start of a request to ensure the check is
made frequently enough.
SPR-8997
Prior to this change, single quotes were incorrectly parsed by
NamedParameterUtils#parseSqlStatement, resulting in incorrect parameter
counts:
ParsedSql sql = NamedParameterUtils
.parseSqlStatement("SELECT 'foo''bar', :xxx FROM DUAL");
assert sql.getTotalParameterCount() == 0 // incorrect, misses :xxx
That is, presence of the single-quoted string caused the parser to
overlook the named parameter :xxx.
This commit fixes the parsing error such that:
ParsedSql sql = NamedParameterUtils
.parseSqlStatement("SELECT 'foo''bar', :xxx FROM DUAL");
assert sql.getTotalParameterCount() == 1 // correct
Issue: SPR-8280
ResourceDatabasePopulator is a component that underlies the database
initialization support within Spring's jdbc: namespace, e.g.:
<jdbc:initialize-database data-source="dataSource">
<jdbc:script execution="INIT" location="classpath:init.sql"/>
</jdbc:initialize-database>
Prior to this commit, ResourceDatabasePopulator#executeSqlScript's use
of Statement#executeUpdate(sql) precluded the possibility of SELECT
statements because returning a result is not permitted by this method
and results in an exception being thrown.
Whether this behavior is a function of the JDBC specification or an
idiosyncracy of certain implementations does not matter as the issue
can be worked around entirely. This commit eliminates use
of #executeUpdate(sql) in favor of #execute(sql) followed by a call
to #getUpdateCount, effectively allowing any kind of SQL statement to
be executed during database initialization.
Issue: SPR-8932
Prior to this commit, StandardServletEnvironment's servlet context
PropertySource remained stubbed out until it the ServletContext became
available and could be replaced during the refresh() of its enclosing
WebApplicationContext. This behavior is acceptable in most cases.
However, if the user has declared an ApplicationContextInitializer that
attempts to access servlet context-params via the Environment API, this
result in a kind of 'false negative', i.e. the context-param key and
value are actually present in the ServletContext, but the PropertySource
representing servlet context params is still a stub at this point,
meaning that it returns an empty result in all cases.
With this change, WebApplicationContextUtils#initServletPropertySources
is invoked eagerly by the ContextLoader if any ACI classes have been
declared. This swaps out the servlet context property source stub for
the real thing just in time for ACIs to use it if necessary.
Extra guard logic has been added to #initServletPropertySources to
ensure idempotency -- once the stub has been replaced, the method
never attempts the replacement again, e.g. during the normal context
refresh() when this method will be called again.
Issue: SPR-8991
- Perform early check whether any ACI classes have been declared and
exit immediately if not, avoiding any other processing
- Polish method names in ContextLoaderTests
Issue: SPR-8991
The spring-aspects Maven pom had an incorrect compile-scoped dependence
on spring-test. In fact, spring-aspects only uses spring-test in its
unit tests. The pom has been updated accordingly, meaning that use
of spring-aspects in Maven-based applications will no longer require
spring-test on the classpath at runtime.
ivy.xml metadata did not need updating, as it was already correct.
This change is only necessary on the 3.1.x line; in 3.2.x (master) Maven
poms are generated automatically from Gradle dependency metadata, which
is also already correct.
Issue: SPR-9048
A list of "known" session attributes (listed in @SessionAttributes)
was gradually built as attributes get added to the model. In a
failover scenario that knowledge is lost causing session attributes
to be potentially re-initialized via @ModelAttribute methods.
With this change @SessionAttributes listed by name are immediately
added to he list of "known" session attributes thus this knowledge
is not lost after a failover. Attributes listed by type however
still must be discovered as they get added to the model.
Prior to this commit, @Configuration classes included via @Import (or
via automatic registration of nested configuration classes) would
always be registered with a generated bean name, regardless of whether
the user had specified a 'value' indicating a customized bean name, e.g.
@Configuration("myConfig")
public class AppConfig { ... }
Now this bean name is propagated as intended in all cases, meaning that
in the example above, the resulting bean definition of type AppConfig
will be named "myConfig" regardless how it was registered with the
container -- directly against the application context, via component
scanning, via @Import, or via automatic registration of nested
configuration classes.
Issue: SPR-9023
A number of users reported issues with comparing method identity vs
equivalence when discovering JavaBeans property methods in
ExtendedBeanInfo.
This commit updates the implementation to consistently use '.equals()'
instead of '=='.
Issue: SPR-8079, SPR-8347
This issue originates from a need in Spring Data JPA, wherein a custom
InstantiationAwareBeanPostProcessor may alter the predicted type of
FactoryBean objects, effectively preventing retrieval of those beans via
calls to #getBeansOfType(FactoryBean.class).
The reason for this "masking effect" is that prior to this change, the
implementation of AbstractBeanFactory#isFactoryBean considered only the
"predicted type" returned from #predictBeanType when evaluating
assignability to FactoryBean.class
The implementation of #isFactoryBean now ensures that not only the
predicted bean type is considered, but also the original bean
definition's beanClass (if one is available).
Issue: SPR-8954
Previously, the build script was configured to add ajbuilder to the set
of Eclipse/STS build commands, meaning that both javabuilder and
ajbuilder would be present for spring-aspects. This causes unpredictable
behavior, as these two builders compete with each other. As ajbuilder is
a functional superset of javabuilder, this commit ensures that only the
former is present for spring-aspects' .project file.
Also removed warning language in import-into-eclipse.sh about
spring-aspects failing after adding Git support, as this intermittent
problem was almost certainly an artifact of the situation described
above.
This change introduces a protected ReflectiveMethodResolver#getMethods,
allowing subclasses to specify additional static methods not
declared directly on the type being evaluated. These methods then become
candidates for filtering by any registered MethodFilters and ultimately
become available within for use within SpEL expressions.
Issue: SPR-9038
User may already have run `gradle build` successfully. This change
ensures that we do not unnecessarily remove these output directories
in order to avoid forcing the user to rebuild.
The "default" FlashMapManager implementation added in 3.1 was invoked
after the redirect, which is too late in cases where the HTTP session
has not been yet been created since as the response is committed.
This change corrects the issue and makes other improvements to the
FlashMapManager implementation such as extracting a base
AbstractFlashMapManager class and making it easier for other
implementations to be added (for example cookie-based).
This is the first merge from 3.1.x => master after the Gradle build
system migration. Notice how files changed under the 3.1.x directory
structure (org.springframework.*) merge seamlessly into the new
directory structure (spring-*).
Certain files had changed under 3.1.x that have since been deleted with
the Gradle build migration, e.g. all pom.xml files had <license>
sections added. These files showed up as a conflict during the merge,
but the resolution is to simply re-remove them and commit as they are
no longer relevant under 3.2.x / master.
Also noteworthy is the .gitignore file. It has been updated under 3.1.x
to ignore files and directories specific to the new Gradle-based
structure. However, this causes conflicts when trying to merge against
master, given that master should *not* ignore this directories. The
resolution in this situation is to simply force the 'master' version of
the file, i.e. when prompted for merge resolution:
anakata:~/Work/spring-framework/spring-framework[master|MERGING]
$ git status -sb
## master...springsource/master [ahead 24]
UU .gitignore
anakata:~/Work/spring-framework/spring-framework[master|MERGING]
$ git checkout master .gitignore
anakata:~/Work/spring-framework/spring-framework[master|MERGING]
$ git commit
It is helpful in situations like this one to enable git's "rerere"
feature beforehand, which records and remembers resolution strategies,
avoiding the need to repeat them in future merges:
$ git config --global rerere.enabled 1
See:
http://progit.org/2010/03/08/rerere.htmlhttp://gitfu.wordpress.com/2008/04/20/git-rerere-rereremember-what-you-did-last-time
Conflicts:
.gitignore
.springframework.*/pom.xml
This merge migrates the Spring Framework 3.2.x build system to Gradle.
Major changes
- Remove Ant-based spring-build and related resources
- Replace with Gradle-based build
- Remove (and .gitignore) all IDE metadata files
- Remove hand-maintained Maven poms in favor of generation by Gradle
- Move integration-tests subproject to root src/test dir
- Move spring-framework-reference subproject to root src/reference dir
- Rename org.springframework.* subprojects => spring-*
See individual messages for the commits included in this merge for
details on each of these changes.
Documentation
- https://github.com/SpringSource/spring-framework#building_from_source
- https://github.com/SpringSource/spring-framework/wiki
see 'Build and release FAQ' and 'SpringSource repository FAQ'
Issue: SPR-8116
When switching back to 3.1.x from master, ignore renamed directories,
Gradle 'build' dirs, generated IDE metadata, etc.
You may wish to clean these files with
$ git clean -dfx
Or do a dry-run beforehand with the '-n' flag:
$ git clean -dfxn
- Update building from source section with Gradle instructions
- Add import-into-eclipse.sh interactive helper script
- Add import-into-idea.md with steps and known issues
Note that use of STS Gradle tooling was attempted, but several issues
remain before it can handle the spring-framework build. In the meantime
the instructions laid out in import-into-eclipse provide an error-free
import.
Eclipse .project, .classpath, and .settings metadata have already been
removed. Now removing Eclipse .psf and formatter XML files from the root
project as well as removing all manually-maintained IDEA .iml and .ipr
files.
This is in favor of using Gradle's own 'eclipse' and 'idea' tasks that
generate this kind of metadata on the fly.
.gitignore has been updated to ignore these files going forward. In any
case, they should not be checked into the source tree! The README.md
file will be updated to explain how to generate these files using gradle
and how to import the projects into your IDE of choice.
Each of these tests began failing during the Gradle build porting
process. None seem severe, many are likely due to classpath issues.
In the case of TestNG support, this needs to be added to the Gradle
build in order to execute these tests. See SPR-8116.txt
- Fix compileTestJava issue in which test classes were not being
compiled or run
- Use built-in eclipse.project DSL instead of withXml closure
to add AspectJ nature and builder
- Rename {aspectJ=>aspects}.gradle and format source
e.g.:
Implementation-Title: spring-core
Implementation-Version: 3.2.0.BUILD-SNAPSHOT
Setting these values is good as a general practice, but required in
order to support the functionality in spring-core's SpringVersion class.
This renaming more intuitively expresses the relationship between
subprojects and the JAR artifacts they produce.
Tracking history across these renames is possible, but it requires
use of the --follow flag to `git log`, for example
$ git log spring-aop/src/main/java/org/springframework/aop/Advisor.java
will show history up until the renaming event, where
$ git log --follow spring-aop/src/main/java/org/springframework/aop/Advisor.java
will show history for all changes to the file, before and after the
renaming.
See http://chrisbeams.com/git-diff-across-renamed-directories
Understanding Gradle pom generation
-------------------------------------------
All spring-* subprojects have had Gradle's 'maven' plugin applied to
them. This means that one can run `gradle install`, and POMs will be
generated according to the metadata in the build.gradle file.
The 'customizePom' routine added by this commit hooks into this
generation process in order to add elements to the pom required for
entry into Maven Central via oss.sonatype.org[1].
This pom generation happens on-the-fly during `gradle install` and
the generated poms exist only in your local .m2 cache. Therefore,
you will not see the poms on the source tree after this command.
Handling optional and provided dependencies
-------------------------------------------
Note particularly the handling of 'optional' and 'provided'
dependencies. Gradle does not have a first class notion for these
concepts, nor are they significant to the actual Gradle build process,
but they are important when publishing POMs for consumption via Maven
Central and other Maven-compatible repositories.
<optional>true</optional> indicates that a dependency need not be
downloaded when resolving artifacts. e.g. spring-context has an
compile-time dependency on cglib, but when a Spring user resolves
spring-context from Maven Central, cglib should *not* automatically
be downloaded at the same time. This is because the core functionality
within spring-context can operate just fine without cglib on the
classpath; it is only if the user chooses explicitly to use certain
functionality, e.g. @Configuration classes, which do require cglib,
that the user must declare an explicit dependency in their own build
script on cglib.
Marking these kinds of dependencies as 'optional' provides a kind of
built in 'documentation' about which version of cglib the user should
declare if in fact he wishes to.
Spring has a great many compile-time dependencies, but in fact very
few mandatory runtime dependencies. Therefore, *most* of Spring's
dependencies are optional.
<scope>provided</scope> is similar to 'optional', in that dependencies
so marked should not be automatically downloaded during dependency
resolution, but indicates rather that they are expected to have been
provided by the user application runtime environment. For example, the
Servlet API is in fact a required runtime dependency for spring-webmvc,
but it is expected that it will be available via the user's servlet
container classpath. Again, it serves here as a kind of 'documentation'
that spring-webmvc does in fact expect the servlet api to be available,
and furthermore which (minimum) version.
This commit adds two closures named 'optional' and 'provided' as well as
two arrays (optionalDeps, providedDeps) for tracking which dependencies
are optional or provided. An optional dependency is declared as follows:
compile("group:artifact:version", optional)
Here, the optional closure accepts the dependency argument implicitly,
and appends it to the 'optionalDeps' array. Then, during pom generation
(again, the customizePom routine), these arrays are interrogated, and
pom <dependency> elements are updated with <optional>true</optional> or
<scope>provided</scope> as appropriate. Thanks to the Spock framework
for inspiration on this approach[2].
[1] http://bit.ly/wauOqP (Sonatype's central sync requirements)
[2] https://github.com/spockframework/spock/blob/groovy-1.7/gradle/publishMaven.gradle#L63
Juergen Hoeller
Rossen Stoyanchev
Chris Beams
Arjen Poutsma
Thomas Risberg
Kris De Volder
Andy Clement