Camel 2.6.0 release
New and NoteworthyWelcome to the 2.6.0 release which approx 297 issues resolved (new features, improvements and bug fixes such as...) - Fixed issue in OSGi with refreshing Camel bundles causing a race condition when discovering type converters
- Introduced ScheduledRoutePolicy with the Quartz component
- Java DSL RouteBuilder now fails if
onException and the likes are not configured before routes, which they must. - Easier to debug routes from
camel-test using the Debugger - Fixed problem installing
camel-cxf feature on Apache Karaf. The EIP now supports property placeholders in the String based options (a few spots in Java DSL where its not possible). For example:
<convertBodyTo type="String" charset="{{foo.myCharset}}"/>
- Introduced ScheduledRoutePolicy to allow you to schedule when routes should be active.
- Introduced
stepwise option to FTP component to allow end users to control how to traverse directories of the FTP servers. This allows you to set stepwise=false to return back to the behavior from Camel 2.0 - 2.4. See more at the FTP documentation. - Added
separator option to FTP component to allow end users to control what path separator to use when uploading files to remote FTP servers. For example if a Windows based FTP server only accept Windows paths, then you can set this option to Windows . - Thread names now outputs CamelContext name which makes it easier to differentiate when running multiple Camel applications in the same JVM.
- Introduced
contentCache option to XSLT component to allow easy reloading of stylesheets during development. - Improved Camel JMX to always find next free name when registering in JMX and existing name exists (to remedy name clash). This makes it easier when deploying muliple WARs in the same JVM where the Camel applications may have same CamelContext id. In OSGi the name in JMX now has the bundle id as prefix.
- Fixed BeanShell language
- Quartz now supports using older versions (eg versions 1.6/1.7) of quartz JARs.
- The Aggregator EIP will use a synchronous invocation when processing completed aggregated Exchanges, when not running in parallel mode. This ensures no internal task queue is used, which otherwise could stack up tasks and under certain conditions eat up memory. Enable
parallelProcessing if you want to use a separate thread for processing completed aggregated Exchanges. - LDAP supports paging.
- CXF and CXFRS use the CXF continuation API when using CXF 2.3.x.
- You can access the CXF Message from the Camel message header by using the key "CamelCxfMessage" after the CXF consumer.
- Fixed a rare situation in which an Exchange could be lost during graceful shutdown on the SEDA endpoints, usually more apparent on high volume systems.
- Improved support for Locale in Bindy for Number and Date formats.
- Fixed Multicast, Splitter, Recipient List to work as the Pipes and Filters would do, breaking to continue routing if a sub exchange failed and was handled by an Exception Clause. Previously you would have to detect this manually in the
AggregationStrategy . Constant fields from Exchange can be referred in the DSL from a String attribute using the class type style. This makes the Java DSL and Spring XML more similar. And you got type safe as Camel throws exception if you had a typo. For example:
<setHeader headerName="Exchange.FILE_NAME"><constant>myfile.txt</constant></setHeader>
- Exchange now contains the creation timestamp when it was created as a
java.util.Date object stores as a property with the key Exchange.CREATION_TIMESTAMP - Camel fails fast on startup if Load Balancer has been misconfigured, such as configuring two different load balancers in one
- Fixed
markRollbackOnlyLast when used with inner/outer transactions. InterceptStrategy is now default ordered in the order they are added to the CamelContext. However you can dictate the order by implementing the org.apache.camel.util.Ordered interface.- Made it easier to load routes from XML files using
loadRoutesDefinition method from CamelContext. - Added support for custom multipart filters in Jetty.
- Using Processor or Bean in routes is now also enlisted in JMX for management.
- The ObjectName for
processor in JMX is now always using the node id it has been assigned in the route. - Using a custom Processor or Bean will be enlisted in JMX having its custom attributes/operations listed. For example just use the Spring JMX annotations to easily expose the attributes/operations. See more details here.
- Fixed some issue with the
camel-blueprint component. - Added features to
camel-blueprint component so its on-par with feature set from Spring XML. - Fixed FTP producer may creating directory using windows path separators.
- Fixed InterceptSendToEndpoint not honoring
stop DSL to stop immediately and not sending the Exchange to the intercepted Endpoint. - Improved the IRC component
- Added
matchsMockWaitTime method to NotifyBuilder which is a convenient method to use when testing with mocks and want to match using timeout. org.apache.camel.spi.EventNotifier events for Exchange done is now fired after OnCompletion has been executed.- Added option
deleteOutputFile to XSLT component. - Idempotent Consumer's
IdempotentRepository is now managed in JMX. - File and FTP producers now support writing done files after the target file has been written, using the new
doneFileName option. - File and FTP consumers now support only consuming files if a done file exists, using the new
doneFileName option. - Thread Pools is now named more intuitive in JMX, making it much easier for end user to spot which pool is used for what. And it provides more information where the pool is used.
- Data Format now supports
Service and CamelContextAware callbacks which means its much easier to perform custom initialization logic in your data formats. - Fixed Printer component to use exact same constants for
mediaSize option as from the Java Printer API. - Fixed issue with route scoped Exception Clause may pick Exception Clause defined on another route, if exception is the same type. Also ensures that route scoped is preferred over context scoped.
- Fixed Bindy fixed length to throw exception if field exceeds allowed length. Added option
clip on @DataField to allow Bindy to clip the field so it fits the length, instead of thrown an exception. - Fixed Bindy when using quoted CSV records and fixed when using pipeline delimiter.
- Added header
Exchange.REDELIVERY_MAX_COUNTER to message when Camel performs a redelivery attempt. This header contains the maximum redelivery setting. It may be null if you use unlimited redelivery attempts or use retryWhile . - JMS producer using
InOnly MEP now supports JMSReplyTo configured on either endpoint, or provided as message header. The message send to the JMS destination will now contain the JMSReplyTo information. The JMS producer which send the message will still be InOnly and not expect/wait for a reply but continue immediately. You have to have the preserveMessageQos=true option defined as well. - Fixed JMS producer, when a
JMSReplyTo header was provided as a message header, removing any queue: or topic: prefix causing the reply to destination to be resolved correctly (eg avoiding having queue: in the queue name). - JMS producer now honors, if reply to was disabled using
disableReplyTo=true in all situations. - HTTP, Jetty, SERVLET and
camel-http4 now supports content type application/x-java-serialized-object on both the producer and consumer side. This allows you to transfer serialized Java objects over http. - Added option
transferException to HTTP, Jetty and SERVLET, which allows you to return back caused exceptions from the consumer side as serialized objects. On the producer side the caused exception will be thrown instead of the generic HttpOperationFailedException . - Web Console now displays some basic route statistics if JMX is enabled.
- JPA
JpaTraceEventMessage now uses @Lob for fields which may contain a lot of data, such as message body. - Throughput logger now supports logging per interval (timer based).
- Fixed an issue with Splitter causing exception mistakenly from one sub-message being propagated to the next sub-message.
- XQuery now strips all whitespaces by default; there is an option to changes this to ignorable whitepsaces only.
- Quartz now enforces trigger name/group to be unique within the given component. Will throw exception on name clash detected. This behavior does not apply for clustered quartz.
- Fixed Splitter using too much memory when running in parallel mode. Now it aggregates on-the-fly and memory consumption is low and stable.
- Bean now has a type option which you can use if you use method overloading in your Bean and prevent Camel to fail with an
AmbiguousMethodCallException . - Added
Main class to camel-core so you can do: Running Camel standalone and have it keep running - Added JDBC based persistent support for Aggregator EIP
- Fixed Jetty may throw NPE during heavy load.
- Continuation timeout can be configured on Jetty. By default Jetty uses 30 sec timeout. There is also a new option to disable using Jetty continuation.
- Polling Consumer's
PollingConsumerPollStrategy now provides number of messages actually processed in the poll in the commit callback message. You can use this to know if none messages was polled, or that X number of message was polled. - Dozer now works with OSGi
- Fixed an issue with Aggregator which could lead to it not confirm completions on the
AggregationRepository . And as well the internal "in progress completed exchanges" map wouldn't neither remove the completed exchange, causing it to grow memory over time. - The
bean DSLs will eager check on startup, the provided method name is a valid method name on the bean as well. This helps to fail fast if end users have a typo in their routes. - Graceful Shutdown a single route now supports a
abortOnTimeout option, which if enabled will let the route be running if it could not shutdown the route within a given timeout period. - Fixed
loadRoutesDefinition and addRouteDefinition to used shared logic from camel-core-xml to ensure loading routes from XML files works exactly the same as if the routes was defined in Spring or Blueprint XML files. - Camel now fail on startup if it cannot load type converters from the classpath, which is essential for it to operate correctly.
- Fixed
completionFromBatchConsumer option on the Aggregator not aggregating the last incoming Exchange under certain conditions. - Fixed a few code generation issues related to Camel Maven Archetypes
- Properties component will trim values from loaded properties. This fixes a problem when you have trailing spaces for values in a properties file.
- Fixed
trim option on @DataField annotation not working in Bindy when unmarshalling fixed length records. - Added
ref function to Simple expression language so you can lookup any Bean from the Registry.
New DSL- Added
excludePattern parameter to removeHeaders DSL.
New AnnotationsAPI breakingcamel-test JAR in the CamelSpringTestSupport classes now returns a more generic AbstractApplicationContext type in the createApplicationContext method.- Added
prepareShutdown method to ShutdownAware SPI interface. org.apache.camel.spi.IdempotentRepository now extends org.apache.camel.Service .- The
onThreadPoolAdd method on LifecycleStrategy has additional parameters for information where the thread pool is used. processBatch method in BatchConsumer now returns number of messages that was actually processed from the batch.commit method in PollingConsumerPollStrategy now has a parameter with the number of message processed in the poll.poll method in ScheduledPollConsumer now returns number of messages that was actually processed from the poll.TypeConverterLoaded has been moved from package org.apache.camel.impl.converter to org.apache.camel.spi .
Known Issues- The Tracer may not output all details for some situations such as when using
onCompletion or intercept etc. - Not all Examples have ANT build.xml files to run the example using ANT.
- Project cannot be fully build using Maven 3.0
- One may encounter build errors in camel-ftp with java versions older than "1.5.0_24"
camel-script which uses ScriptEngineManager from the JDK, cannot load script engines from packages outside the JDK (i.e. when running in OSGi).- Stopping a route after using
adviceWith maybe cause issues with error handlers, see CAMEL-3534 - If you use Quartz then pay attention that its UpdateChecker is enabled by default. You should disable this, see more details here
- Installing
camel-ftp feature in Apache ServiceMix may cause issue. See here for more details and workaround. - Splitter in parallel mode may block if empty body as input. See CAMEL-3685
- Recipient List in parallel mode doesn't reuse background aggregator thread. See CAMEL-3727
- The
camel-spring-ws feature does not work in Apache ServiceMix or Apache Karaf.
Important changes to consider when upgrading- Upgraded to Spring 3.0.5
- Upgraded to Apache CXF 2.3.1 (although CXF 2.2.11 is still supported)
- Upgraded to Jetty 7.2.2.v20101205
- The Java DSL now enforces
onException and the likes to be defined before routes, if not Camel will fail starting the route. - The Aggregator EIP will use a synchronous invocation when processing completed aggregated Exchanges, when not running in parallel mode. This ensures no internal task queue is used, which otherwise could stack up tasks and under certain conditions eat up memory. Enable
parallelProcessing if you want to use a separate thread for processing completed aggregated Exchanges. - Camel now fails fast if staring a
LifecycleStrategy fails on startup. For example if there is issue with JMX that will now cause Camel to not startup at all. If you have trouble with JMX before you can disable JMX on Camel and allow it to startup. ManagementAware has been @deprecated. Just enrich your class with the Spring JMX annotations, if you want custom management of your components, endpoints, processors and so fort. See more at Why is my processor not showing up in JConsole.- Data Format now supports
Service and CamelContextAware callbacks which means its much easier to perform custom initialization logic in your data formats. - The
mediaSize option on Printer now uses exact same constants as the Java Printer API. Which means it uses underscores instead of hyphens in the constants. - JPA
JpaTraceEventMessage now uses @Lob for fields which may contain a lot of data, such as message body. - Quartz now enforces trigger name/group to be unique within the given component. Will throw exception on name clash detected. This behavior does not apply for clustered quartz.
- Camel now fail on startup if it cannot load type converters from the classpath, which is essential for it to operate correctly.
Getting the DistributionsBinary DistributionsSource DistributionsGetting the Binaries using Maven 2To use this release in your maven project, the proper dependency configuration that you should use in your Maven POM is:
<dependency>
<groupId>org.apache.camel</groupId>
<artifactId>camel-core</artifactId>
<version>2.6.0</version>
</dependency>
SVN Tag Checkout
svn co http://svn.apache.org/repos/asf/camel/tags/camel-2.6.0
ChangelogFor a more detailed view of new features and bug fixes, see:
|
SearchError rendering macro 'html' Your Confluence administrator has disallowed the use of Javascript in the HTML macro. This setting can be changed using HTML for Confluence Configuration. Please see your administrator for details. Apache Software Foundation
|