Since SpringFactoriesLoader.loadFactoryNames() will be deprecated in
gh-27954, this commit removes the use of it in the spring-test module.
Specifically, this commit removes the protected
getDefaultTestExecutionListenerClasses() and
getDefaultTestExecutionListenerClassNames() methods from
AbstractTestContextBootstrapper and replaces them with a new protected
getDefaultTestExecutionListeners() method that makes use of new APIs
introduced in SpringFactoriesLoader for 6.0.
Third-party subclasses of AbstractTestContextBootstrapper that have
overridden or used getDefaultTestExecutionListenerClasses() or
getDefaultTestExecutionListenerClassNames() will therefore need to
migrate to getDefaultTestExecutionListeners() in Spring Framework 6.0.
Closes gh-28666
This commit leverages a subset of @philwebb initial experimentation
to compute at build time the value of specific boolean static fields
in native images. This enhancement is implemented for now as a
GraalVM feature.
The goal here is to keep an optimized footprint via build time code
removal without leveraging build-time class initialization which is known
for the blocking compatibility issues it introduces due to its viral nature.
For now, the static fields initialized at build time with native are:
- NativeDetector#imageCode
- Fields with a name ending by "Present" in "org.springframework" package
typically used for classpath check with ClassUtils#isPresent
Closes gh-28624
ProblemDetail is intended to be extended with additional fields. This
commit removes its "with" methods for chained initialization to keep
it as plain as possible and avoid imposing a particular style on
subclasses.
See gh-27052
Ideally one would pass WebClient directly to HttpServiceProxyFactory,
but two need to remain decoupled. This commit adds static, shortcut
methods to WebClientAdapter to create an HttpServiceProxyFactory, thus
eliminating the step to wrap the WebClient.
This commit moves the responsibility of naming classes to the
GenerationContext. This was already largely the case before, except that
the concept of a "mainTarget" and "featureNamePrefix" was specific
to bean factory initialization contributors.
ClassNameGenerator should now be instantiated with a default target
and an optional feature name prefix. As a result, it does no longer
generate class names in the "__" package.
GeneratedClasses can now provide a new, unique, GeneratedClass or
offer a container for retrieving the same GeneratedClass based on an
identifier. This lets all contributors use this facility rather than
creating JavaFile manually. This also means that ClassNameGenerator
is no longer exposed.
Because the naming conventions are now part of the GenerationContext, it
is required to be able to retrieve a specialized version of it if a
code generation round needs to use different naming conventions. A new
withName method has been added to that effect.
Closes gh-28585
This commit refines BindingReflectionHintsRegistrar to skip
registration of hints for Object.class, primitive types and
skip members for array types.
Closes gh-28683
Align IllegalStateException with SQLException handling and propagate
the original exception.
See gh-28669
Co-authored-by: Christoph Mies <chr.mi@web.de>
Prior to this commit, Spring MVC and Spring WebFlux would not support
conditional requests with `If-Match` preconditions. As underlined in the
RFC9110 Section 13.1, those are related to the `If-None-Match`
conditions, but this time only performing requests if the resource
matches the given ETag.
This feature, and in general the `"*"` request Etag, are generally
useful to prevent "lost updates" when performing a POST/PUT request: we
want to ensure that we're updating a version with a known version or
create a new resource only if it doesn't exist already.
This commit adds `If-Match` conditional requests support and ensures
that both `If-Match` and `If-None-Match` work well with `"*"` request
ETags.
We can't rely on `checkNotModified(null)`, as the compiler can't decide
between method variants accepting an ETag `String` or a Last Modified
`long`. Instead, developers should use empty ETags `""` to signal that
no resource is known on the server side.
Closes gh-24881
Since LoggingCacheErrorHandler was only recently introduced in 5.3.16,
we have decided to completely revise its internals (protected API) in
5.3.x while retaining the current public API.
Specifically, this commit:
- introduces protected getLogger() and isLogStackTraces() methods to
improve extensibility
- revises logCacheError() to accept a Supplier<String> for lazy
resolution of error messages
Closes gh-28672
See gh-28670, gh-28648
At present, creating a LoggingCacheErrorHandler that logs stack traces
also requires supplying the logger to be used. This might be
inconvenient for some users, as it requires usage of the Commons
Logging API.
This commit simplifies creation of such as LoggingCacheErrorHandler
instance by adding a constructor that only accepts a boolean flag
indicating whether to log stack traces.
Closes gh-28670
When an overloaded method accepts additional "optional" arguments, we
typically declare the optional arguments after the required arguments.
For example, see the constructors for ClassPathResource.
This commit therefore revises the signature of the overloaded
forResourceLocation() method so that the optional ClassLoader argument
follows the required `String resourceLocation` argument.
* pr/28664:
Polish "Fix enclosing class in TypeReference for inner type arrays"
Fix enclosing class in TypeReference for inner type arrays
Closes gh-28664
Sam Brannen
rstoyanchev
Sébastien Deleuze
Arjen Poutsma
Stephane Nicoll
Kevin Schönfeld
Brian Clozel
Vedran Pavic
Moritz Halbritter
Jerome Prinet
Johnny Lim