CompositeRetryListener currently uses LinkedList to store registered
listeners. The primary operation on this list is iteration (traversing
all listeners on every retry lifecycle event). ArrayList provides
better iteration performance due to better cache locality and lower
memory overhead.
Closes gh-37231
Signed-off-by: Chengang Guan <guanchengang@qq.com>
This commit introduces additional unit tests for CallMetaDataContext's
function return parameter matching in reconcileParameters(), verifying
that the declared return parameter is correctly resolved regardless of
whether it is declared before or after an additional OUT parameter.
See gh-37206
CallMetaDataContext.reconcileParameters() keys the map of declared
parameters by lowerCase(provider.parameterNameToUse(name)), but the
branch that matches the return parameter reported by the database
metadata did not apply the same rule. It looked up the function return
name as declared (original case) and fell back to the first declared
OUT parameter name with a plain toLowerCase(), without the provider
transformation that strips the '@' prefix on SQL Server and Sybase.
The first lookup therefore always missed on Oracle, so the fallback
silently used whichever OUT parameter was declared first. Declaring an
additional OUT parameter before the return parameter of a function made
that parameter double as the return slot: the declared return parameter
was dropped from the call parameters, the wrong parameter was bound at
position 1, and executeFunction() returned the value of the other out
parameter. On SQL Server, a procedure compiled with withReturnValue()
and an '@'-prefixed OUT parameter declared before the return parameter
failed with InvalidDataAccessApiUsageException because neither lookup
could find the declared parameter.
The return parameter branch now looks up the metadata-derived name
first and normalizes both the function return name and the first OUT
parameter fallback with the same rule as the declared parameter map.
Tests cover both declaration orders for an Oracle function and for a
SQL Server procedure with a return value.
Closes gh-37206
Signed-off-by: junhyeong9812 <pickjog@gmail.com>
As a follow up to 1b56f58999, this commit introduces
hasUnbalancedBrackets() in PropertyAccessorUtils, which
AbstractNestablePropertyAccessor uses to reject property paths with
unbalanced '[' or ']' brackets by throwing a
NotReadablePropertyException with an informative message, thereby
improving diagnostics for users.
See gh-36999
Most of the documentation updates were already done in gh-36803, this
completes the section with some information on the property path syntax
supported by allowFields/disallowFields.
Closes gh-36789
Prior to this commit, the reconnection attempt count provided to the
reconnect strategy would not give the updated, incremented value but
instead the previous one.
This commit fixes this and ensures the value is incremented before it's
given to the strategy.
Fixes gh-37223
Prior to this commit, the `DefaultClientResponseBuilder` would assume
that an original client HTTP response, when mutated, would not be reused
nor read anymore. While this is the advised use case, there was some
inconsistency with the builder API here when mutating: some data like
the response status would copied, but the HTTP headers and cookies would
refer directly to the previous entries, making all changes visible to
the previous response instance.
This commit ensures that deep copies are performed when mutating a
client response with the builder API.
Fixes gh-37086
MIME type parameter names are case-insensitive, and MimeType already
stores them in a LinkedCaseInsensitiveMap. Several code paths, however,
still compared them with case-sensitive String.equals().
As a result, MimeType.hashCode() disagreed with MimeType.equals() for
parameter names that differ only in case, breaking the equals/hashCode
contract: text/plain;FOO=bar and text/plain;foo=bar are equal but hash
differently, so one is not found in a hash-based collection holding the
other. MimeType.compareTo() had the same blind spot for the charset
parameter.
MediaType was affected in two further ways: an out-of-range quality
value escaped validation when spelled Q=, and removeQualityValue() left
a Q= parameter in place.
Signed-off-by: Artyom Tsvirko <36863599+lArtiquel@users.noreply.github.com>
FlashMap.startExpirationPeriod(int) and MockMvcWebConnection's cookie
handling both multiplied an int number of seconds by 1000 without
widening to long. Above 2_147_483 seconds (about 24.9 days) the
multiplication overflows to a negative offset, so the computed
expiration time lands in the past.
For FlashMap, a flash map configured through
AbstractFlashMapManager.setFlashMapTimeout(int) with a large timeout is
then treated as expired immediately. For MockMvcWebConnection, a cookie
with a large max-age is removed from the CookieManager instead of being
stored.
This applies the same widening already used for this pattern in
gh-25613.
Closes gh-37208
Signed-off-by: kogun <akogun@gmail.com>
MIME type parameter names are case-insensitive, but MimeTypeParser
accumulates parameters in a case-sensitive LinkedHashMap. As a result,
duplicate parameters differing only in case (such as "charset" and
"CHARSET") were not rejected and were silently collapsed to the last
value by the case-insensitive parameter map of MimeType.
Accumulate parameters in a LinkedCaseInsensitiveMap so that duplicates
differing only in case map to the same key and are rejected consistently
with exact duplicates.
Signed-off-by: junhyeong9812 <pickjog@gmail.com>
Co-authored-by: Yash <190389954+yashsiwacha@users.noreply.github.com>
This commit picks up where 97067f9c9c left off by removing the four
trailing collection-clearing calls in clear(), which are now redundant
since the removal loop already empties contextMap, hierarchyMap,
contextUsageMap, and unusedContexts as an invariant. This commit also
adds regression tests in LruContextCacheTests for closing a context
hierarchy via clear() and reset(), asserting bottom-up close order with
Mockito.inOrder(), and confirming that getParentContextCount() and
getContextUsageCount() return to zero.
See gh-36825
This updates the TestContext Framework cache so that clearing the
cache also closes cached ConfigurableApplicationContext instances
instead of only dropping the internal references. The implementation
reuses the existing removal path, preserving the hierarchy-aware close
behavior already used by cache eviction/removal.
The ContextCache contract now documents the close behavior for
clear(), and LruContextCacheTests covers both clear() and reset(),
since reset() delegates to clear().
See gh-26196
Closes gh-36825
Signed-off-by: Will-thom <116388885+Will-thom@users.noreply.github.com>
Previously, canonicalPropertyName() located the end of a [key]
expression via a naive indexOf("]") search, while
AbstractNestablePropertyAccessor.getPropertyNameKeyEnd() -- used during
actual property resolution -- tracked bracket nesting depth. This meant
the two methods could disagree on the canonical form of a property
path whose map key itself contains bracket characters (e.g.,
map['key[0]']).
Similarly, getNestedPropertySeparatorIndex() tracked whether a dot
separator occurs inside a [key] expression using a simple boolean
toggle that flips on both '[' and ']', which produces an incorrect
result when a key contains an odd net count of inner bracket
characters.
To address those inconsistencies, this commit extracts the private
getPropertyNameKeyEnd() method from AbstractNestablePropertyAccessor to
a package-private static utility method in PropertyAccessorUtils, so
that canonicalPropertyName() can reuse the same depth-aware bracket
matching, and getNestedPropertySeparatorIndex() has been reworked to
track bracket nesting depth instead of toggling a boolean flag.
Closes gh-36999
Prior to this commit, the `DefaultApiVersionStrategy` reactive variant
would attempt to resolve the API version with the first resolver that
replies with a version. This contradicts the API that registers
resolvers in order.
This commit ensures that each resolver is called in order.
Signed-off-by: Gimin Kim <138752849+Gimini-3@users.noreply.github.com>
HtmlUnit 5 moved its Cookie type to org.htmlunit.http, so
Spring Test's HtmlUnit integration now uses the new API.
Constraint: Keep HtmlUnit and htmlunit3-driver versions compatible.
Rejected: Upgrade HtmlUnit alone | driver requires HtmlUnit 5.4.0.
Confidence: high
Scope-risk: narrow
Directive: Keep HtmlUnit and the driver version aligned.
Tested: JAVA_HOME=/opt/homebrew/opt/openjdk@25 ./gradlew build
Not-tested: None
Signed-off-by: jungh8n <jh981113@naver.com>
BodyState.flush() emits every queued buffer and only clears the queue
afterwards, so a cancellation arriving while it emits makes dispose()
release buffers whose ownership has already been transferred to the sink.
Such a buffer is then released twice: once by the parser, and once by the
downstream consumer or the discard hook. With Netty, body buffers are
slices of the inbound buffer, so the second release frees the inbound
buffer prematurely, which surfaces as
IllegalReferenceCountException: refCnt: 0, decrement: 1
io.netty.handler.codec.http.DefaultHttpContent.release
reactor.netty.channel.FluxReceive.drainReceiver
when reactor-netty releases its own share right after onNext.
Remove each buffer from the queue before emitting it, mirroring what
enqueue() already does, so that dispose() only ever releases buffers the
parser still owns.
Signed-off-by: Hyunsik Kang <cj848@hanmail.net>
When a multipart subscriber cancels while MultipartParser has already
emitted body tokens beyond the downstream demand, those tokens are held
in the Flux.create sink queue (and in downstream operator queues such
as windowUntil). On cancellation, Reactor discards the queued tokens,
but BodyToken is not a DataBuffer, so the buffers inside the discarded
tokens are never released and Netty reports "LEAK: ByteBuf.release()
was not called before it's garbage-collected".
Register a doOnDiscard hook for BodyToken in MultipartParser.parse() so
that a discarded body token releases its buffer, both in the sink queue
and in any downstream operator queue that supports discarding.
Closes gh-37115
Signed-off-by: Hyunsik Kang <cj848@hanmail.net>
Sam Brannen
Chengang Guan
Hyunwoo Jung
junhyeong9812
Brian Clozel
rstoyanchev
Juergen Hoeller
Artyom Tsvirko
kogun
Yanming Zhou
Will-thom
Manu Sridharan
Gimin Kim
jungh8n
Hyunsik Kang