Collection of Spring Batch Examples, covering the basics, readers, writers, listeners and complex usecases.
Spring Batch Examples exist, because i needed a central place to store the source code of my own examples collection. Instead of the usual private Subversion repository i wanted to give Github a try.
The Examples are described in each of the submodule README.
This repository uses git submodules to link to the individual example repositories on github. Right after the check-out you need to run: git submodule --init
to get the submodule sources.
To get the latest commits for the submodules:
- update submodules from with git command:
git submodule -q foreach git pull -q origin master
- sourcetree: just fetch/pull from the current repository, sourcetree will update the submodules too
All Spring Batch Examples:
- are individual github repositories and maven projects, the pom.xml in this root directory is only for a convenient build all feature
- are tested with:
- Spring Batch 2.1.8.RELEASE
- Spring Framework 3.1.0.RELEASE
- are provided "as is", no guarantees :-)
- work with in-memory database, but not with in-memory jobrepository, since the MapJobRepository is threadsafe i could use it, but why break a standard configuration ?
Overview for the project setup.
The pom.xml in this root directory contains just a simple parent pom wrapper project to provide a convenient "build all" feature, see Maven Pom project aggregation. Each individual project stands on its own and can be used as such, e.g. there are no maven configurations made in the parent pom.
The examples modules are:
- spring batch examples parent, the mentioned "build all" parent module
- complex, contains examples which use more than one of the core aspects of spring batch
- listeners
- playground, mostly for incubating new examples
- readers
- writers
- -DreuseForks
- run tests with individual and fresh JVM per test class or reuse JVM between tests
- default is
false
- use
true
, if you want to run the tests with shared JVMs
For each project i added specific build configurations for the following build plugins:
- maven-compiler-plugin
- JDK set to 1.6
- compiler.debug=true
- compiler.optimize=false
- maven-enforcer-plugin
- enforce minimum Java version 1.6
- enforce minimum Maven version 3.0
- maven-resources-plugin
- forced UTF-8 encoding
- maven-surefire-plugin
- set log4j properties file location
- Java memory configuration to prevent OutOfMemory problems during tests, see Java -X options
- forkCount=1C to run one test class per cpu core
- reuseForks=false to run each test class isolated
- can be overridden by using systemproperty
-DreuseForks=true
- can be overridden by using systemproperty
I use the reuseForks=false for the maven test configuration to get a new JVM for each test class. This configuration lowers the test run time by a signifikant amount. Why is that necessary?
While using the HSQLDB in-memory database i see sometimes thread problems.
Each project contains only the needed dependencies and i check regularly for version updates and the [dependencies licenses][license].
The project follows the maven standard directory layout, only difference so far is a README.md (md for markdown format) and a LICENSE file instead of both files ending with .txt.
Overview:
# log4j configuration
src/main/resources/log4j/log4.properties
# the job configurations
src/main/resources/spring/batch/job/
# spring batch setup
src/main/resources/spring/batch/setup/job-context.xml
src/main/resources/spring/batch/setup/job-database.xml
# the used input files
src/test/resources/input/
# test setup
src/test/resources/spring/batch/setup/test/job-test-context.xml
For each project:
- the log4j.properties is under
src/main/resources/log4j/log4j.properties
- logging level is WARN for all and DEBUG for the source package of the project
- location might be changed soon to src/test...
- the Spring Batch infrastructure setup is under
src/main/resources/spring/batch/setup/...
job-context.xml
contains JobRepository, StepScope etc. beansjob-database.xml
contains the datasource and transactionmanager beans- HSQLDB in-memory variant is used
- Spring Batch Schema is loaded at Application Context startup with jdbc:initialize-database
- the Spring Batch test infrastructure setup is under
src/test/resources/spring/batch/setup/test/...
job-test-context.xml
contains just the JobLauncherTestUtils bean
- each example has its own package (test package has the same name), e.g.
simplelist
- not all examples have java source, some have only a job.xml and some tests
- each example has its own job.xml, e.g.
simple-list-job.xml
- each example has a large test coverage, well what can i say, i am addicted to tests :-)
To simplify it, all work is under Apache 2.0 license, fork it, use it, bend it and if you find a bug or improvement, do not hesitate to push a patch.
last check: 29.04.2013
- AOP Alliance - Public Domain
- Apache commons-collections, -dbcp, -io, -logging, -pool - all licensed under Apache 2.0
- bcprov - MIT
- harmcrest-core - new BSD
- HSQLDB - based on BSD
- Jettison - Apache 2.0 license
- log4j - Apache 2.0 license
- Maven - Apache 2.0 license
- slf4j - identical to MIT License
- Spring Batch - Apache 2.0 license
- Spring Framework - all licensed under Apache 2.0
- xpp3 - XMLPullParser - Public Domain
- xstream - BSD
- JUnit - Common Public License - v 1.0 (look for license.txt in the github repository)
- just do not distribute your project with junit, should be easy
- Truezip - Eclipse Public License, Version 1.0
- if you distribute a project with truezip you need to include the license and a statement basically saying you did not change anything
- if you change the source, ... well if you run a commercial product you are screwed :-)