This commit compiles our Protobuf against 4.27, effectively raising our
baseline to 3.9+.
This commit also re-generates all the Java messages from the .proto spec
using the latest protoc binary.
Closes gh-33011
This commit harmonizes the equals/hashCode behavior of OverrideMetadata
to always take the implementation class as a factor for its identity.
This is important as two OverrideMetadata implementations could use
the same strategy and other settings while creating the override value
in a totally different way. This commit makes sure they are identified
as different.
Closes gh-33005
This commit harmonizes how a candidate bean definition is determined
for overriding using `@TestBean`, `@MockitoBean`, and `@MockitoSpyBean`.
Previously, a qualifier was necessary even if the name of the annotated
field matches the name of a candidate. After this commit, such candidate
will be picked up transparently, the same it is done for regular
autowiring.
This commit also reviews the documentation of the feature as considering
the field means that its name is taken into account to compute a cache
key if by-type lookup is requested.
Closes gh-32939
This commit fixes a package tangle between the root bean override
package and its sub-packages. This was vastly improved with the
introduction of `@DummyBean` but we still had a few tests that were
at the wrong place.
This commit changes how factory method for `@TestBean` usage is
discovered. Previously the field name or bean name suffixed with
'TestOverride' was used. It sounds more natural to just use the
field name or bean name, leaving cases where a suffix is required
to explicitly providing the method name.
As part of this change, the exception messages have been revisited as
it's less since the method name candidates have the exact same name
as the field or bean name. A `()` is added to make it more clear the
name is for a method.
Closes gh-32940
TestOverrideMetadata was doing byName lookup by default, even without
a bean name. This makes writing tests that need to do by type lookup
way more complicated that it should.
This commit introduces DummyBean, that merely replaces two types with
hardcoded values. It also exposes the raw attribute of OverrideMetadata
and doesn't override anything from it to make sure the behavior is not
altered.
Closes gh-32982
This commit aligns the by-type lookup for bean overrides with what was
done when a bean name is present. It now correctly generate a bean
name rather than failing because no bean of that type exists.
Closes gh-32990
Prior to this commit, the `ContentCachingRequestWrapper` could allocate
a `FastByteArrayOutputStream` block that was larger than the content
cache limit given as a consturctor argument. This was due to an
optimization applied in gh-31834 for allocating the right content cache
size when the request size is known.
Fixes gh-32987
By swallowing the exception on failing to release a savepoint, we are hiding
client code from database errors that they may need to be aware of or handle.
e.g. if the connection is now dead when trying to release the savepoint, by
swallowing the exception here, client code has no way of knowing this now until
the next operation on the database, which will yield a confusing message about
the problem and obscures where the failure occurred as it will be in a different
part of their code that throws any exception.
Swallowing the exception here is also inconsistent with all the other database
calls in the Transaction Object, where we always raise an exception up and let
client code decide how to handle the error.
Prior to this commit, the `ServerHttpObservationFilter` would support
async dispatches and would do the following:
1. start the observation
2. call the filter chain
3. if async has started, do nothing
4. if not in async mode, stop the observation
This behavior would effectively rely on Async implementations to
complete and dispatch the request back to the container for an async
dispatch. This is what Spring web frameworks do and guarantee.
Some implementations complete the async request but do not dispatch
back; as a result, observations could leak as they are never stopped.
This commit changes the support of async requests. The filter now
opts-out of async dispatches - the filter will not be called for those
anymore. Instead, if the application started async mode during the
initial container dispatch, the filter will register an AsyncListener to
be notified of the outcome of the async handling.
Fixes gh-32730
This commit reviews how bean override support can influence the key of
an application context cached by the TCF. OverrideMetadata and its
subclasses now implement a proper equals/hashCode pair that is tested
in various scenarios.
Due to how the TCF operates, OverrideMetadata has to be computed in
two locations:
1. In a ContextCustomizerFactory, using the metadata in the enclosing
class if any. This determines whether a customizer is needed in the
first place. The computed set of unique metadata identifies the
customizer and participates in the application context cache's key.
2. In the TestExecutionListener so that it knows the override points
it has to process.
Parsing of the metadata based on a test class has been greatly
simplified and moved to OverrideMetadata proper as we don't need several
flavors. 1 and 2 are using the same algorithm with the former wrapping
that in a Set to compute a proper key.
BeanOverrideContextCustomizerEqualityTests provides a framework for
testing edge cases as we only care about whether the created
ContextCustomizer behaves correctly against the identity of another.
Closes gh-32884
This commit reviews the structure of several classes to comply with
our guidelines. Also, rather than exposing a static method to configure
the context in a test, we call the high-level API directly.
Rather than having to override the getBeanName method when there is one,
this commit moves it as a @Nullable argument. This makes sure that
equals and hashCode consistently use the bean name. getBeanName cannot
be final just yet as the test infrastructure overrides it.
Also, arguments are now ordered consistently, which improves code
readability and type signature.
Commit d7970e4ab8 introduced support for JAXBElement in
Jaxb2XmlEncoder's encodeValue() method; however, canEncode() still
returned false for a JAXBElement element type.
This commit revises canEncode() so that it returns true for an element
type that is assignable from JAXBElement.
See gh-30552
See gh-32972
Closes gh-32977
Stéphane Nicoll
Brian Clozel
Sébastien Deleuze
Juergen Hoeller
rstoyanchev
hlmg
Carlos Corbacho
Matteo Cristoforo
Sam Brannen
Thomas Deblock