I recently wrote a blog post[1] sharing my experiences with using Apache Spark to load data into Apache Fluo. One of the things I cover in this blog post is late binding of dependencies and exclusion of provided dependencies when building a shaded jar. When writing the post, I was unsure about dependency isolation and convergence expectations in the Spark env.
Does Spark support any form of dependency isolation for user code? For example can the Spark framework use Guava ver X while user code uses Guava version Y? This is assuming the user packaged Guava version Y in their shaded jar. Or, are Spark users expected to converge their user dependency versions with those used by Spark? For example, the user is expected to converge their code to use Guava version X which is used by the Spark framework. [1]: http://fluo.apache.org/blog/2016/12/22/spark-load/ --------------------------------------------------------------------- To unsubscribe e-mail: user-unsubscr...@spark.apache.org
