Guava 11.0.2 downgrade; Improvements for reduce-side joins

By changing the centipede dependence we roll back to the Guava 11.0.2 packaged with Hadoop to avoid "dependency hell". Numerous small changes have been made to the reduce-side join code to help with telepath work.

Bump to centipede 99.3, Spring 4, Guava 16

If you hate XML, now you can configure infovore with Groovy

Persistent AWS Clusters

In this release, Haruhi can create a persistent cluster in AWS that does not terminate when it runs out of steps. Haruhi can now submit jobs to a running persistent cluster, which reduces the turnaround in testing small jobs from about ten minutes to two minutes.

Foreach loop for Haruhi flows; support for telepath development

This release includes a foreach loop for Haruhi flows, which makes it easy to run a number of similar jobs in sequence. This release also includes a number of refactorings and other changes to support the creation of projects like telepath, which derive from bakemono.

Stronger Unit Testing For Reducers, Centipede and Spring Version Bump

Toughen up unit testing #96 which replicates the behavior of Hadoop, and bump Centipede version and Spring versions #97

Extract Facts About Subjects Of Specified Types

This edition contains #92 (Tool to extract objects with specified types) and the supporting change #95 to add two cluster types with increased memory, necessary for running the new extractIsA tool.

This edition has been used to create skibase an extract of information about Ski Areas.

rdfDiff

This release introduces a new general purpose tool rdfDiff that can be used to take the delta of two sets of triples in RDF Patch format. (See #82)

This release also configures logging in Haruhi to suppress the torrent of INFO messages generated by the Spring framework (See #90)

Bump centipede version, use classpath scanning to find apps

To prepare for applications with complex options, I bumped the Centipede version to 98.7, also I've switched the Main app to use Spring and classpath scanning to find applications, which takes a step out of creating an application and means new apps can be added by just adding classes to the classpath.

t20131125 -- create sieve3 streams for "literals" and "dotdot" predicates

This release changes the organization of :BaseKB Now in that literals go into a file called literals #86 and predicates with a double dot (which are said to be erroneous) go into the dotdot file #87. Internally work has continued on the use of the set difference operator to validate Freebase data #83 and #70. See the t20131125 milestone for the list of changes.

t20131118

Major changes include #79 (set difference operator) and certain work towards #70 (extraction of unique predicates, subjects and objects) as well as #78, which sets the correct jar versions for the "haruhi" script and makes haruhi look for the same version of bakemono as itself.

A full list of changes is here