splout-hadoop MR2 via Maven dependency downloads Pangool MR1 transitively

[pereferrera]

Because of the way Pangool is deployed to Maven, splout-hadoop downloads always the base artifact (without -mr2) even when depending on splout-hadoop with classifier = mr2. This is a bug. Right now affected users need to exclude the pangool dependency and add their own like in this example:

<dependency>
  <groupId>com.datasalt.pangool</groupId>
  <artifactId>pangool-core</artifactId>
  <version>0.60.3</version>
  <classifier>mr2</classifier>
</dependency>
<dependency>
  <groupId>com.splout.db</groupId>
  <artifactId>splout-hadoop</artifactId>
  <version>0.2.5</version>
  <classifier>mr2</classifier>
  <exclusions>
    <exclusion>
      <groupId>com.datasalt.pangool</groupId>
      <artifactId>pangool-core</artifactId>
    </exclusion>
  </exclusions>
</dependency>

In the future we should either define pangool-core as scope provided and force the users to declare pangool explicitly or find another solution.