Prior to this commit, the exception message generated by
instantiateFactory() in SpringFactoriesLoader did not report the
offending class in the top-level exception message. The offending class
was in fact included in the message of the nested exception, but the
top-level exception message on its own was a bit misleading.
This commit improves the diagnostics for such failures by including the
offending class and the target factory type in the top-level exception
message.
Closes gh-22453
Before this commit, the return type for Kotlin suspending functions
(as returned by MethodParameter#getParameterType and
MethodParameter#getGenericReturnType methods) was incorrect.
This change leverages Kotlin reflection instead of Java one
to return the correct type.
Closes gh-21058
- Integration tests run with zero copy configuration.
- RSocketBufferLeakTests has been added.
- Updates in MessagingRSocket to ensure proper release
See gh-21987
Prior to this commit, repeated calls to `DataBuffer.write(CharSequence,
Charset)` would not write the given chars to the expected position in
the original buffer.
This commit fixes that by writing to the `outBuffer.position()`, offset
by the current `DataBuffer.writePosition()`.
Fixes gh-22484
Prior to this commit, calls to `MimeType` and `MediaType` would create a
significant amount of garbage:
* during startup time, in the static sections of `MimeType` and
`MediaType` when creating well-known types
* at runtime, when parsing media types for content negotiation or
writing known media types as strings in HTTP response headers
This commit does the following:
* Avoid parsing the well-known types and use regular constructors
instead
* Cache types in a simple LRU cache once they've been parsed, since an
application is likely to deal with a limited set of types
* Avoid using `java.util.stream.Stream` in hot code paths
Benchmarks show that a complete revision of the `MimeTypeUtils` parser
is not required, since the LRU cache is enough there.
Closes gh-22340
Prior to this commit, profiling sessions would show that using
`java.util.stream.Stream` in some hot code paths creates significant
garbage.
Where streams aren't really required, this commit turns those snippets
into imperative logic because those are likely to be called once or
multiple times per request.
Closes gh-22341
Prior to this commit, `DataBuffer.write` would throw an
`IllegalStateException` when called with an empty `String`, since the
`CharsetEncoder` would be flushed on an incorrent state.
This commit skips entirely the encoding phase for empty `String`.
Fixes#22262
This allows for proper nested type conversion in @Value Optional fields analogous to method parameters, through a new TypeDescriptor-based method in the TypeConverter SPI. As an additional and less involved measure that is worth backporting, DefaultListableBeanFactory defensively checks for pre-converted Optional wrappers.
Issue: SPR-17607
Aalto's InputFactoryImpl already disables loading of external entities
by default (property "javax.xml.stream.isSupportingExternalEntities").
This commit goes further by applying the same defensive measures as we
do elsewhere for XMLInputFactory, which disables DTD completely.
Arguably there is no good reason to enable that by default in WebFlux.
Typically a straight up equals as well as Collections#contains
checks for MediaType.ALL is susceptible to the presence of
media type parameters.
This commits adds equalsTypeAndSubtype as well as an
isPresentIn(Collection<MimeType>) methods to MimeType to faciliate
with checks for MediaType.ALL.
Issue: SPR-17550
Prior to this commit, one could write a `CharSequence` to an existing
`DataBuffer` instance by turning it into a byte array or `ByteBuffer`
first. This had the following disadvantages:
1. Memory allocation was not efficient (not leveraging pooled memory
when available)
2. Dealing with `CharsetEncoder` is not always easy
3. `DataBuffer` implementations, like `NettyDataBuffer` can use
optimized implementations in some cases
This commit adds a new `DataBuffer#write(CharSequence, Charset)` method
for those cases and also an `ensureCapacity` method useful for checking
that the current buffer has enough capacity to write to it..
Issue: SPR-17558
Prior to this commit, parallel invocations of
MethodParameter.findParameterIndex() (invoked indirectly via
SynthesizingMethodParameter.forParameter() and
MethodParameter.forParameter()) could intermittently lead to an
IllegalArgumentException being thrown due to a race condition in the
internal implementation of the JDK's
java.lang.reflect.Executable.getParameters() method.
This commit addresses this issue by introducing a fallback for-loop
that iterates over the candidate parameters a second time using
equality checks instead of identity checks.
Issue: SPR-17534
This commit reverts the first optimizations listed in
fa096dc60f, as the default delimiters
do vary, namely by the charset given in the message mime type.
The mimetype charset might not be compatible with ASCII (i.e. anything
but UTF-8 or ISO-8859-1, for instance it might be UTF-16), and will not
successfully find the default delimiters as a consequence.
Added test to indicate the bug.
This CGLIB upgrade literally does not make any difference to Spring users since its only purpose is a default ASM7 declaration in its AsmApi class (which we patched a month ago). We nevertheless leave our local AsmApi copy in place for the time being, allowing for an individual upgrade of the declared ASM API version in the future.
Issue: SPR-17267
Issue: SPR-17371
1. Avoid re-creating the List with delimited byte arrays on every
request if using the default delimiters which don't vary by charset.
2. Replace flatMap with flatMapIterable for splitOnDelimiter.
3. Avoid going through DataBufferUtils#join, and unnecessarily creating
Flux from the List, since the join method needs a list anyway.
Fix ResourceRegionEncoder so that it checks for resource existance
before writing boundaries. Also defer data buffer allocation until
necessary.
Issue: SPR-17419
As discussed in KT-25165, from a Kotlin POV enum constructors
have no parameter, this is an "implementation detail"
required for running on the JVM, so it seems relevant to skip
Kotlin reflection in that case and just delegate to Java
reflection.
Issue: SPR-16931
Includes use of Files.getLastModifiedTime for NIO Paths, preservation of NIO-based resolution on createRelative, deprecation of PathResource, and consistent use of getContentLengthLong over getContentLength.
Issue: SPR-17320
This was a package private class in spring-messaging since 5.0, and was
recently made public in 5.1. This commit promotes it to spring-core
where it belongs next to all other ListenableFuture support classes.
Follow-up refactoring for SPR-17336
Previously, both nullsLow(Comparator<T>) and
nullsHigh(Comparator<T>) returned a
NullSafeComparator which treated nulls as being
high. This commit corrects this typo.
Prefix the opening brace with a single space in some classes.
Replace tab character with space in javadoc of AttributeAccessor.
Also fix some other trivial formatting errors.
Closes gh-1979
1. Helper method to eliminate duplication in formatting (de-)serialized
values for logging introduced with prior commit #e62298.
2. Helper method for TRACE vs DEBUG logging with different details.
Issue: SPR-17254
At DEBUG show up to 100 chars, at TRACE show full formatted value.
Note that the formatValue helper method is duplicated a number of times
in this commit. A utility method will likely be added in spring-core
through an extra commit.
Issue: SPR-17254
This commit makes sure that in DataBufferUtils.write, any received data
buffers are returned as part of the returned flux, even when an error
occurs or is received.
Issue: SPR-16782
Before this commit, Kotlin inner class nested configuration
handling thrown an IndexOutOfBoundsException due to bogus filtering
of its constructor parameter reference to an instance of the outer
class.
This commit keep constructor parameter of type INSTANCE in order to
throw a more meaningful NoSuchBeanDefinitionException.
Issue: SPR-17222
This commit adds decoder/message-reader tests for errors in
the source data buffer publisher. Because the tests extend
AbstractDataBufferAllocatingTestCase, they also check whether
the buffers that precede the error in the stream are properly
released.
Issue: SPR-17025
Includes caching of declared annotation arrays and combined searching for several annotation types (used in SpringCacheAnnotationParser).
Issue: SPR-16933
This revision considers any java/javax and org.springframework.lang annotations as not synthesizable upfront, checking not only in isSynthesizable but also at synthesizeAnnotation(Array) level.
Issue: SPR-16933
Empty Maps are preferably initialized without capacity (not initializing them at all or lazily initializing with default capacity when needed).
Issue: SPR-17105
This commit includes an optimization of BeansUtils#instantiateClass
that favors Java reflection for default constructors before leveraging
Kotlin one for finding primary constructors and avoids Kotlin related
conditions when running in Java.
Issue: SPR-17069
This revision limits serializability of derived interfaces, superclasses and type parameters, optimizing for introspection performance instead.
Issue: SPR-17070
Includes tests for parsing all available locales on the JVM, checking toString/toLanguageTag equality between parsed and original locale.
Issue: SPR-7598
Issue: SPR-16651
Like CookieLocaleResolver, LocaleChangeInterceptor parses both locale formats by default now. Since it does not need to render the locale, its languageTagCompliant property is not relevant anymore at all.
The parseLocale method in StringUtils validates the locale value now and turns an empty locale into null, compatible with parseLocaleString behavior and in particular aligned with web locale parsing needs.
Issue: SPR-16700
Issue: SPR-16651
Introduces a configure method pattern for Supplier-style configuration and a common SingletonSupplier decorator for method reference suppliers. Also declares jcache.config and jcache.interceptor for non-null conventions.
Issue: SPR-17021
In order to allow Spring Framework applications running as GraalVM
native images, ReactiveAdapterRegistry should perform explicit class
checks instead of catching Throwable in order to avoid
UnsupportedFeatureError errors.
Issue: SPR-17000
Includes an efficient implementation of isEmpty(), not relying on a full entry count but rather backing out once a non-empty hash segment has been found.
Issue: SPR-16994
Polish a few issue identified when adding checkstyle to the
build. Although checkstyle is not enforcing rules on tests,
these are a few minor changes that are still worth making.
Issue: SPR-16968
Reorganize imports to ensure consistent ordering. This commit also
expands any `.*` static imports in favor of using fully-qualified
method references.
Issue: SPR-16968
Update all classes so that inner classes are always last. Also
ensure that utility classes are always final and have a private
constructor and make exceptions final whenever possible.
Issue: SPR-16968
Allow the `Environment` to accept a generic `Profiles` interface which
can support more complex matching rules. The previous
`acceptsProfiles(String...)` method now uses `Profiles.of` which
supports basic profile expressions such as "(a | b) & !c"
Issue: SPR-12458
Prior to this commit, the generated POMs for Spring Framework modules
would contain unneeded/harmful information from the Spring Framework
build:
1. The BOM imports applied to each module by the dependency
management plugin, for example for Netty or Reactor Netty.
Spring should not export that opinion to its POMs.
2. The exclusion of "org.slf4:jcl-over-slf4j" from *all* dependencies,
which made the POMs much larger than necessary and suggested to
developers that they should exclude it as well when using all those
listed dependencies. In fact, only Apache Tiles currently brings that
transitively.
This commit removes that information from the POMs.
The dependencyManagement Gradle plugin is disabled for POM generation
and we manually resolve the dependency versions during the generation
phase.
The Gradle build is streamlined to exclude "org.slf4:jcl-over-slf4j"
only when necessary.
Issue: SPR-16893
Includes efficient check for same ClassLoader in ClassUtils.isVisible, efficient MethodMatchers check for IntroductionAwareMethodMatcher, and supertype method resolution in MethodMapTransactionAttributeSource.
Issue: SPR-16723
Spring's CGLIB fork is patched with local copies of affected files here, introducing the notion of a "contextClass" (e.g. the proxy superclass) which gets passed through to ReflectUtils.defineClass for delegating to MethodHandles.Lookup.defineClass eventually, against a privateLookupIn(contextClass) lookup context on JDK 9/10/11.
Issue: SPR-15859
This commit introduces tests for looking up annotations on parameters
in constructors for nested and inner classes via Spring's
MethodParameter abstraction.
The test for an inner class is currently disabled since it fails on
JDK 8. See JIRA issue for details.
Issue: SPR-16652
The prevailing current wisdom is to use the default constructor for
secure and let it pick the best algorithm for the OS.
On Java 8 (Oracle), Linux this results in "NativePRNG" which uses
/dev/random (potentially blocking) for the initial seed, and
/dev/urandom (non-blocking) for subsequent calls to nextInt.
Issue: SPR-16635
stsypanov
Sam Brannen
Konrad Kamiński
Rossen Stoyanchev
Brian Clozel
Andrew Tulloch
Juergen Hoeller
Sebastien Deleuze
olszewskimichal
Arjen Poutsma
Andrzej Leśkiewicz
igor-suhorukov
Sam Brannen
Vojtech Janota
Stephane Nicoll
Hanope
Dimitrios Liapis
Sam Hubbard
www
Marten Deinum
Sergey Chupov
Kazuhiro Sera
Phillip Webb
Yun
Oleksandr Hasheniuk
nkjackzhang
Andreas Asplund
Kazuki MATSUDA