I've been using the %dep extensively to setup my notebook dependencies for later use by the Scala Spark paragraphs.
My main use case for these notebooks involve the following steps - Spin up an ephemeral cluster (with Spark and Zeppelin on it) - Import the notebook - Run I've seen https://issues.apache.org/jira/browse/ZEPPELIN-630 and the comments on https://zeppelin.apache.org/docs/0.6.1/interpreter/spark.html#dependencyloading which seem to suggest that this feature in being deprecated in favor of the GUI settings since it works for both %spark and %sql interpreters. >From a user's point-of-view the way to fixing this should've been fixing the loading for %sql or at least warning about it so it doesn't surprise users. But to takeaway something that is already working well, and remove it in favour of matching a the less powerful %sql approach? -- Cheers, Ahmed http://bit.ly/ahmed_abtme <http://about.me/humanzz>