v1.2.1

This release adds a :lazy option to the output DSL method, which configures LazyOutputFormat.

v1.2.0

• New DSL syntax for defining whether jobs should run in parallel or sequentially. You can now describe complex workflows with parallel and sequential steps, see #29
• It's now easier to make sure that a job gets the input that a previous job wrote, see #26
• Support for custom input formats
• Support for using Java classes for mappers, reducers and most of the other configurations that take a class
• JRuby 1.6.x is no longer supported

Besides new features the test setup has been overhauled to make it easier to add new tests and a way to run integration style tests without shelling out to Hadoop. The tests are also no longer dependent on RVM.

This release was possible only because of @grddev and @jowl, thank you for your contributions!

v1.1.3

Bugfix release

• No longer packages the *.class files twice into rubydoop.jar
• Correctly calls #setup on combiners.

Thanks to @bjorne & @grddev for this release

v1.1.2

Compatibility with newer versions of JRuby (v1.7.6 and above) that bundle some default gems in the distribution.

v1.1.1

Bugfix release. Fixes a race condition where the proxy layer could accidentally create a mapper object instead of a reducer object for a combiner. This should only have happened when running Hadoop in local mode with a combiner.

v1.1.0

• Fixes performance issues with combiners by using a single global runtime per JVM instead of a runtime per processor (mapper, reducer, combiner, etc.)
• The JRuby embedding code has been updated to be future proof
• Improved the gem packaging method to be more compatible with gems like jruby-openssl
• Support for YARN/Hadoop 2.0