https://jira.springsource.org/browse/INT-3146
Incompatibility of using socket timeouts with a caching client
connection factory.
When a socket option timeout (soTimeout) is set on a TCP connection
and the timeout occurs, the socket is closed.
When a connection is intercepted, the close is performed through
the interceptor (for example to allow a closing handshake before the
physical close).
However, the caching client connection factory is implemented
using an interceptor, which returns the underlying connection to
the cache pool (for reuse). In the case of a TcpNetConnection,
the reader thread has terminated meaning that, the next time the connection
is used, no reply will ever be received.
The work-around (when using a gateway) is to use the 'remote-timeout'
attribute instead of relying on the soTimeout. There is no
work around when using collaborating channel adapters with a Net
connection. Using NIO works because there is no reader thread
in that case, but the socket is never closed on a timeout.
Always physically close the connection whenever an exception occurs
even if the close was delegated to an interceptor.
Add test cases for Net and NIO implementations.
* Add XSD attributeGroup `smartLifeCycleAttributeGroup`
* Use it for all adapters tags
* Refactoring `AbstractChannelAdapterParser` hierarchy
* Refactoring some `MessageHandler`'s that should not have `autoStartup` & `phase`.
It is the responsibility of `AbstrctEndpoint`
* Polishing some tests according new logic
JIRA: https://jira.springsource.org/browse/INT-2998
INT-2998: Polishing
Polishing
- Reformat schemas where indentation changed
- Add SLC to the UDP inbound adapter
Previously, `RequestHandlerRetryAdvice` wrapped a Handler's (business) Exceptions in a
`MessagingException` on each retry within `RetryCallback#doWithRetry`
and before `RetryPolicy#canRetry`.
It made useless some retry framework out-of-the-box features like `BinaryExceptionClassifier`
* Push wrapping to `MessagingException` after `retryTemplate#execute`
* Polishing `ErrorMessageSendingRecoverer` regarding new logic
* Add `retryableExceptions` test
You can now properly specify Business exceptions with the retry policy to
perform conditional retry.
JIRA: https://jira.springsource.org/browse/INT-2943
Polishing
Only wrap Throwable in ThrowableHolderException if the Throwable
is not an Exception.
INT-2943: Polishing ErrorMessageSendingRecoverer
* Add `<script>` and `<expression>` to `<int:inbound-channel-adapter>`
* Refactoring for `DefaultInboundChannelAdapterParser`
* Introduce `ScriptExecutingMessageSource` to allow scripts' `MessageProcessor`
implementations work as a source for polling endpoint
JIRA: https://jira.springsource.org/browse/INT-2867
INT-2867: documentation
In addition the tests for Lifecycle attributes
Polishing
* #jsonPath function is registered automatically if the library is on the classpath
* Add otional dependency on `com.jayway.jsonpath:json-path` artifact
* Add `JsonPathUtils` class as a wrapper around `JsonPath`
* Add `#jsonPath` tests for: `<transformer>`, `<filter>`, `<splitter>`, `<router>`
JIRA: https://jira.springsource.org/browse/INT-3139
INT-3139: Further improvements
* Remove `SpelFunctionRegistrar`
* Introduce `SpelFunctionFactoryBean`
INT-3139: Polishing
Previously, provided `MessageProcessor` to create a `MessageHandler`
for endpoint was wrapped to `MethodInvokingMessageProcessor`,
e.g. `GroovyScriptExecutingMessageProcessor`.
Check the type of provided object in the constructor
of `MethodInvoking*` strategies and use it directly if it is `MessageProcessor`
JIRA: https://jira.springsource.org/browse/INT-2433
* Add `local-filename-generator-expression` attribute to FTP and SFTP Outbound Gateway
* Refactoring for `(S)FtpOutboundGatewayParser`
* Add `local-filename-generator-expression` tests
* Add 'What's New' and attribute description
JIRA: https://jira.springsource.org/browse/INT-3129
Polishing
Doc Polishing
Assert attribute is only set for get/mget
Move tests to the get gateways
* Convert `row-mapper` attribute of stored procedure components
to bean reference
* Class name is also supported for backward compatibility
JIRA: https://jira.springsource.org/browse/INT-2739
* `AbstractMappingMessageRouter#addChannelFromString` uses
`StringUtils.tokenizeToStringArray`, so it's enough to check it from test.
JIRA: https://jira.springsource.org/browse/INT-2117
`StringHttpMessageConverter` adds `Accept-Charset` header by default.
By RFC 2616 `Accept-Charset` is a Request header,
so it won't be presented in the response.
Turn off `writeAcceptCharset` in the `HttpRequestHandlingEndpointSupport`.
JIRA: https://jira.springsource.org/browse/INT-3120
`DelayerHandlerRescheduleIntegrationTests` fail sometimes
on asserts. Looks like `context.destroy()` works very slow
and `<delayer>` manages to send delayed Messages
JIRA: https://jira.springsource.org/browse/INT-3138
Send is deemed to be successful if sent to at least this
number of subscribers (default 0).
Note: when using a task executor, if there is at least one
subscriber, a send is always good, regardless of success
or failure of invoking the handler.
Polishing - Docs - PR Comments
* add `<spel-function>` XSD element
* add `SpelFunctionParser`
* add `SpelFunctionRegistrar` to avoid introducing some confused 'Method'-bean
* add `SpelFunctionRegistrar` collaboration with `IntegrationEvaluationContextFactoryBean`
* some refactoring for `IntegrationEvaluationContextFactoryBean`
* polishing some failed tests after this change
JIRA: https://jira.springsource.org/browse/INT-1639
INT-1639: SpelFunctionParser use BeanClassLoader
INT-1639: Inherit 'functions' from parent AC
INT-1639: Document the <spel-function>
INT-1639: 'NoSuchBeanDefinitionException' fix
Polishing
Invoke ARPMH Directly (SA, Splitter etc)
If a bean referenced by a service activator etc, is already an ARPMH,
then wire it directly into the endpoint rather than wrapping it in a
MethodInvoking Handler.
Disallow the same bean being used in this way in multiple endpoints.
It would cause rewiring of outputChannel etc.
Detect such use and suggest using prototype scope.
Previously, an ARPMH used in a MethodInvokingHandler didn't work
in some cases, because the "inner" ARPMH had no output-channel
and the outer ARPMH (e.g. transformer) wouldn't get a reply and
would complain if requiresReply is set.
The general solution is to detect the ARPMH targetObject in
AbstractStandardMessageHandlerFactoryBean, invoke canBeUsedDirect()
to allow the subclass to vote - specifically it will be allowed
if the type is not also a special type understood by the subclass
and no endpoint-specific attributes are set.
To this end, remove a few default attributes from the schema to
enable detection of explicit setting of these attributes.
If these conditions pass and the method is handleRequest (either
explicitly or by omission) then wire it in and invoke the
subclass to post process the handler (set attributes).
Restructure this post processing in each subclass so that it can be invoked in
this way, or directly by the subclass when it makes a custom
handler.
Polishing - PR Comments
Mainly a little refactoring in the RouterFactoryBean to allow
the parameter in canBeUsedDirect() to be an ARPMH instead of
Object.
Plus a couple more minor changes.
The `DelayHandler.DelayedMessageWrapper` works correctly, as designed by INT-3049.
However `DelayHandler.DelayedMessageWrapper` contains `Message` as a property
and `GenericMessage` can't be reconstructed automatically.
So, added new `DBObjectToGenericMessageConverter` to read `GenericMessage` for
`DelayHandler.DelayedMessageWrapper#original` property.
JIRA: https://jira.springsource.org/browse/INT-3070
Cherry-picked 2.2.x
INT-3070: PR comments
INT-3070: Remove `assert` for common state
Add an MBeanTreePollingMessageSource that produces a graph of simple objects representing the JMX tree (INT-3124).
The DefaultMBeanObjectConverter converts MBean objects into a graph of Lists, Maps and arrays or primitives.
Formatting tidy, per feedback.
Feedback incorporated: split setter/attributes, more tests, inner bean constructor, logging changes.
Overloaded setter methods with different parameter types are now named separately, the endpoint attributes reflect this and tests are added to reflect this.
An inner bean can be supplied to provide an alternative MBeanObjectConverter.
Log a warning instead of a more destructive UnsupportedOperationException where there's incomplete parsing in the DefaultMBeanObjectConverter and add a trace level for exception debugging.
Minor test change and doc/reference update
Added attribute filter interface as suggested. Three implementations are provided 'all', 'named only' and 'not named'
actually add notnamedfield filter (doh).
Polishing
Polishing tests to follow with new changes in the SPR 4.0
Polishing build.gradle to be compatible with SPR 4.0
Build passes with SPR 3.2.4 too
JIRA: https://jira.springsource.org/browse/INT-3131
Requires Jackson2
Bump Jackson2 version
Remove write support
Add String support
Polishing
Initialize objectMapper field; assert ObjectMapper is not null.
Newlines at ends of files.
Some handlers get a componentName, but not a beanName. For example
the TransformerFactoryBean may create a Transformer that's not
a bean, to wrap a POJO, or expression. The AbstractSimpleMessageHandlerFactoryBean
transfers the component name to the transformer but it has a null
beanName. It would not be appropriate for the FB to change the beanName.
The IntegrationMBeanExporter's fallback, when it can't find the handler
as part of a stand-alone endpoint, is to use the handler's toString().
IntegrationObjectSupport (IOS) toString delegates to Object.toString()
if there is no beanName.
IOS's getComponentName() falls back to the beanName if there is no
component name.
Change the IMBE to use the componentName as the first fallback and,
if null, fallback to the current fallback.
One side effect is that handlers that do have a bean name
will no longer have '.handler' at the end of their name; but this
is not really an issue because the MBean is designated as a
handler anyway (via the bean component of the ObjectName).
However, identifying chain components is new in 3.0 so this is
not really a breaking change.
Previously, 'If-Modified-Since' and 'If-Unmodified-Since' HTTP headers were incorrectly processed
within from/to HTTP headers mapping, because Spring-Web `HttpHeaders` has a confusing method name, see:
https://jira.springsource.org/browse/SPR-10600
* fix 'If-Modified-Since' and 'If-Unmodified-Since' processing independently from `HttpHeaders`
* add fallback to formatted string for date ware HTTP headers
* add Spring Integration HTTP proxy scenario test
JIRA: https://jira.springsource.org/browse/INT-2995
The second chance NIO logic that detects a close being received
in a race condition with a reply needs to be enabled always because
a Failover or Caching connection factory might be using NIO
underneath.
Also, add an epoch to the FailoverTcpConnection's connectionId which
is incremented on each use to prevent a close after a previous
use being received by the current user.
Finally, add code in the FailoverTcpConnection to not propagate
messages from an old (defunct) delegate connection - if the
delegate has changed, the connection must be being used by
another client.
Move definition of the IntegrationEvaluationContextFactoryBean
from DefaultConfiguringBeanFactoryPostProcessor to
AbstractIntegrationNamespaceHandler.
We need a separate eval contexts for each app context so that child
contexts get the correct BeanResolver.
Add Test For IECFB and Parent Context
There were several "helper" classes where the context was initialized
in the constructor, before the BeanFactory was passed in.
Generally fixed by adding the BeanFactory to the constructor args.
Other cases where the container-managed bean instantiated a helper
and never passed in the BeanFactory.
Finally, a fix to ExpressionUtils where the caller had a BeanFactory
but the BF did not contain an EvaluationContext factory bean, the
BeanResolver was not set up. This is unlikely in a Spring Integration
application, but added for completeness.
INT-3115 Add a BeanFactory to Test Cases
Change the WARN log in ExpressionUtils to a fatal exception to detect
cases where an EvaluationContext was created without a BeanFactory.
While this was generally in test cases, it also exposed some cases
in code where the context was initialized without a BF.
polishing on merge
