metron-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From mattf-horton <>
Subject [GitHub] metron pull request #481: METRON-322 Global Batching and Flushing
Date Wed, 12 Jul 2017 21:51:16 GMT
Github user mattf-horton commented on a diff in the pull request:
    --- Diff: metron-platform/metron-writer/src/main/java/org/apache/metron/writer/bolt/
    @@ -92,24 +177,58 @@ public void prepare(Map stormConf, TopologyContext context, OutputCollector
           configurationTransformation = x -> x;
         try {
    -      bulkMessageWriter.init(stormConf,
    -                             context,
    -                             configurationTransformation.apply(new IndexingWriterConfiguration(bulkMessageWriter.getName(),
    -                             getConfigurations()))
    -                            );
    +      WriterConfiguration writerconf = configurationTransformation.apply(
    +              new IndexingWriterConfiguration(bulkMessageWriter.getName(), getConfigurations()));
    +      if (defaultBatchTimeout == 0) {
    +        //This means getComponentConfiguration was never called to initialize defaultBatchTimeout,
    +        //probably because we are in a unit test scenario.  So calculate it here.
    +        BatchTimeoutHelper timeoutHelper = new BatchTimeoutHelper(writerconf::getAllConfiguredTimeouts,
    +        defaultBatchTimeout = timeoutHelper.getDefaultBatchTimeout();
    +      }
    +      writerComponent.setDefaultBatchTimeout(defaultBatchTimeout);
    +      bulkMessageWriter.init(stormConf, context, writerconf);
         } catch (Exception e) {
           throw new RuntimeException(e);
    +  /**
    +   * Used only for unit testing.
    +   */
    --- End diff --
    Hm.  With respect, this one I don't really agree with.  It's just a polymorphism of the
regular prepare() method.  Since its signature differs from the "regular" prepare() method,
system code will never call it.  It is documented, at javadoc level, as being only for use
in unit testing, which is by definition code-level.  Those who aren't willing to look at code
shouldn't be using it, and indeed will have a difficult time finding it at all.  I'd rather
let the polymorphism be evident.
    I will, however, add further comment that this is "used only for unit testing, with injection
of a FakeClock for mocking passage of time in a controlled way, to test queue timeout behavior
in the WriterComponent."  Would that be okay?

If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at or file a JIRA ticket
with INFRA.

View raw message