Prior to this commit, the ApiDiff Gradle task would be configured for
all submodules of the Spring Framework project and would assume that
they all existed for the baseline version considered for the API diff.
This would cause issues when:
* the sub-project is not published as it's not an official "spring-*"
module
* the "spring-*" module is new and did not exist for the baseline
version
This commit ensures that only "spring-*" modules are considered for this
task and that we trigger an early resolution of the baseline version -
if the version doesn't exist, a warn message is logged and we assume
that this is a new module, to be compared with an empty configuration.
This commit also renames the "spring-core-graalvm" project to
"graalvm-feature", since this sub-project is not an official module
published to Maven Central, but rather an internal dependency.
Fixes gh-28818
This commit deprecates ListenableFuture in favor of CompletableFuture.
ListenableFuture was introduced in Spring Framework 4.0, when
CompletableFuture was not yet available. Spring now requires JDK 17, so
having our own type no longer seems necessary.
Major changes in this commit include:
- Deprecation of ListenableFuture and related types
(ListenableFutureCallback, SettableListenableFuture, etc.)
- Deprecation of AsyncListenableTaskExecutor in favor of default methods
in AsyncTaskExecutor (submitCompletable).
- AsyncHandlerMethodReturnValueHandler now has toCompletableFuture
instead of toListenableFuture.
- WebSocketClient now has execute methods, which do the same as
doHandshake, but return CompletableFutures (cf. the reactive
WebSocketClient).
All other changes
- add an overloaded method that takes a CompletableFuture parameter
instead of ListenableFuture, and/or
- add a method with a 'Async' suffix that returns a CompletableFuture
instead of a ListenableFuture (connectAsync, sendAsync).
Closes gh-27780
This commit polishes DefaultGenerationContext to make the method
that flushes generated classes more explicit. It now throws an
IOException and TestGenerationContext has been updated to handle
that to ease its use in code that can't throw such an exception.
As this use case is likely to happen outside the Spring Framework,
this commit adds such a convenience to spring-test as well.
Closes gh-28877
Migrate all AOT tests to make use of `GeneratedClasses` rather than
directly generating Java files. This commit also refines and polishes
AOT APIs to being greater consistency.
Specifically:
- The `MethodGenerator` interface has been removed in favor of
working directly with `GeneratedMethods`.
- The visibility of several constructors and methods has been
reduced to package-private.
- The `using(...)` and `builder` methods have been removed in
favor of setting the `Consumer` callbacks directly as
constructor arguments.
- Variable names for builders are now named `type` or `method`
depending on what they're building.
Closes gh-28831
Prior to this commit, when users wished to register proxy hints for a
Spring AOP JDK dynamic proxy, they were required to explicitly specify
SpringProxy, Advised, and DecoratingProxy along with user interfaces.
This commit simplifies hint registration for Spring AOP proxies by
introducing two completeJdkProxyInterfaces() methods in AopProxyUtils,
one that accepts strings and one that accepts classes that represent
the user-specified interfaces implemented the user component to be
proxied. The SpringProxy, Advised, and DecoratingProxy interfaces are
appended to the user-specified interfaces and returned as the complete
set of interfaces that the proxy will implement.
Closes gh-28745
Since users might not have a concrete need to work with TypeReference,
this commit introduces ProxyHints.registerJdkProxy(String...) to
simplify use of the API for registering a dynamic proxy based on fully
qualified class names of the required interfaces.
Closes gh-28781
The warning message logged for an annotation that still uses
convention-based overrides now includes a concrete suggestion for how
the problem may potentially be fixed.
WARN - Support for convention-based annotation attribute overrides is
deprecated and will be removed in Spring Framework 6.1. Please
annotate the 'locations' attribute in
@org.springframework.core.annotation.MergedAnnotationsTests$ConventionBasedComposedContextConfiguration
with an appropriate @AliasFor declaration -- for example,
@AliasFor(annotation = org.springframework.core.annotation.MergedAnnotationsTests$ContextConfiguration.class).
See gh-28760
This commit replaces convention-based annotation attribute overrides in
tests with explicit use of @AliasFor -- except for tests in spring-core,
since we still want to test our support for convention-based annotation
attribute overrides.
See gh-28760
Implicit convention-based annotation attribute overrides have been
supported for a long time; however, Spring Framework 4.2 introduced
support for explicit annotation attribute overrides via @AliasFor.
Since explicit overrides are favorable to implicit overrides, and since
the support for convention-based overrides increases the complexity of
Spring's annotation search algorithms, this commit deprecates
convention-based overrides and logs a WARNING whey the are encountered.
For example, the following message is logged for a test that still
uses convention-based overrides. A log message is generated for each
such attribute in a given annotation but only once per application run.
WARN Support for convention-based annotation attribute overrides is
deprecated and will be removed in Spring Framework 6.1. Please
annotate the 'basePackages' attribute in
@org.springframework.context.annotation.ConfigurationClassPostProcessorTests$ComposedConfigurationWithAttributeOverrides
with an appropriate @AliasFor declaration.
Closes gh-28760
AnnotationTypeMapping.addConventionMappings() sometimes adds
convention-based mappings that it should not.
For example, in certain circumstances an explicit annotation attribute
override configured via @AliasFor can be incorrectly mapped as
convention-based.
Although this does not appear to cause negative side effects (other
than unnecessary processing), this is technically a bug, and this
commit address this issue.
Closes gh-28773
Prior to this commit, a DataSize input string could not be parsed if it
contained any whitespace.
With this commit, a DataSize input string can contain leading, trailing,
or 'in between' whitespace. For example, the following will be parsed
to the same DataSize value.
- "1024B"
- "1024 B"
- " 1024B "
- " 1024 B "
Closes gh-28643
Brian Clozel
Stephane Nicoll
Arjen Poutsma
Phillip Webb
rstoyanchev
Marc Wrobel
Sam Brannen
Juergen Hoeller
Sébastien Deleuze