Updated READ.me, tweak to make Beam Window act as a Dummy

README.md

@@ -18,9 +18,8 @@ Note you need the Pentaho settings.xml in your ~/.m2 : https://github.com/pentah
 
 * Create a new directory called kettle-beam in \<PDI-DIR\>plugins/ 
 * Copy target/kettle-beam-<version>.jar to \<PDI-DIR\>/plugins/kettle-beam/
-* Copy the other jar files in target/lib to \<PDI-DIR\>/lib
+* Copy the other jar files in target/lib to \<PDI-DIR\>/plugins/kettle-beam/lib/
 
-  I know it's dirty, fixing it later
 
 ## Configure
 
@@ -44,10 +43,24 @@ You can use the variables to make your transformations completely generic.  For
 
 ## Supported
 
-* Only straight line between Beam Input and Output is supported.  
-* Sort and Group by are not yet supported.
-* Group By step : experimental, SUM (Integer, Number), COUNT
-* External plugins are not yet supported.
+* Input: Beam Input and GCP Pub/Sub Subscribe
+* Output: Beam Output and GCP Pub/Sub Publish
+* Windowing with the Beam Window step 
+* Sort rows is not yet supported and will never be supported in a generic sense.
+* Group By step : experimental, SUM (Integer, Number), COUNT, MIN, MAX, FIRST (throws errors for not-supported stuff)
+* Merge Join
+* Stream Lookup (side loading data)
+* Filter rows (including targeting steps for true/false)
+* Switch/Case
+* Plugin support through the Beam Job Configuration: specify which plugins to include in the runtime
+
+## Runners
+* Beam Direct : working
+* Google Cloud DataFlow : working
+* Apache Spark : mostly untested, configurable (feedback welcome)
+* Apache Flink : not started yet, stubbed out code
+* Aache Apex : not started yet, stubbed out code
+* JStorm : not started yet
 
 ## More information
 

src/main/java/org/kettle/beam/steps/window/BeamWindow.java

@@ -8,30 +8,16 @@
 import org.pentaho.di.trans.step.StepInterface;
 import org.pentaho.di.trans.step.StepMeta;
 import org.pentaho.di.trans.step.StepMetaInterface;
+import org.pentaho.di.trans.steps.dummytrans.DummyTrans;
 
-public class BeamWindow extends BaseStep implements StepInterface {
+/** Behaves like a Dummy step, simply passing rows in the regular runner.
+ *  In Beam this code isn't used.  So the data passing in Dummy only is useful for unit testing and development in Spoon.
+ *
+ */
+public class BeamWindow extends DummyTrans implements StepInterface {
 
-  /**
-   * This is the base step that forms that basis for all steps. You can derive from this class to implement your own
-   * steps.
-   *
-   * @param stepMeta          The StepMeta object to run.
-   * @param stepDataInterface the data object to store temporary data, database connections, caches, result sets,
-   *                          hashtables etc.
-   * @param copyNr            The copynumber for this step.
-   * @param transMeta         The TransInfo of which the step stepMeta is part of.
-   * @param trans             The (running) transformation to obtain information shared among the steps.
-   */
-  public BeamWindow( StepMeta stepMeta, StepDataInterface stepDataInterface, int copyNr, TransMeta transMeta,
-                     Trans trans ) {
+  public BeamWindow( StepMeta stepMeta, StepDataInterface stepDataInterface, int copyNr, TransMeta transMeta, Trans trans ) {
     super( stepMeta, stepDataInterface, copyNr, transMeta, trans );
   }
 
-  @Override public boolean processRow( StepMetaInterface smi, StepDataInterface sdi ) throws KettleException {
-
-    // Outside of a Beam Runner this step doesn't actually do anything, it's just metadata
-    // This step gets converted into Beam API calls in a pipeline
-    //
-    return false;
-  }
 }