Thank you Sourav and Amos for the quick replies.  Very helpful.  Will play
around with rZeppelin some more - so far it's been working very well for
our use cases.



On Thu, Dec 24, 2015 at 11:12 AM, Sourav Mazumder <
sourav.mazumde...@gmail.com> wrote:

> In my experience parameter passing between R, scala, Spark SQL SparkR
> worked great in rZeppelin.
>
> I could successfully build a model using SparkR's GLM algorithm, done a
> prediction using the model, then I could create a SparkR datafarme from the
> predicted dataset. Then used the same to visualize the data using R's
> visualization library. I used rZeppelin for the same.
>
> Regards,
> Sourav
>
>
>
> On Thu, Dec 24, 2015 at 11:01 AM, Amos B. Elberg <amos.elb...@me.com>
> wrote:
>
>> Yeah, I’ve looked into it a little bit :p
>>
>> rZeppelin (the one in repository Zeppelin-With-R) is an accepted pull
>> request that is pending integration with Zeppelin.
>>
>> In an rZeppelin session, R shares the same Spark backend with the rest of
>> the Zeppelin Spark-based interpreters.  That’s why multiple languages can
>> share the same Spark data pipeline.
>>
>> Datalayer’s R interpreter is not integrated with the rest of Zeppelin.
>> It launches its own instance of Spark which is separate and distinct from
>> the one being used by the rest of Zeppelin.
>>
>> rZeppelin also supports moving data between languages using the Zeppelin
>> Context, which allows you to share variables with scala and python.
>> Datalayer has their own implementation for moving variables between R and
>> scala.
>>
>> rZeppelin supports interactive visualizations.  I don’t believe that
>> Datalayer’s does.
>>
>> There are other differences as well.  This is explained in the rZeppelin
>> readme.
>>
>>
>> From: Girish Reddy <gir...@springml.com> <gir...@springml.com>
>> Reply: users@zeppelin.incubator.apache.org
>> <users@zeppelin.incubator.apache.org>
>> <users@zeppelin.incubator.apache.org>
>> Date: December 24, 2015 at 1:52:11 PM
>> To: users@zeppelin.incubator.apache.org
>> <users@zeppelin.incubator.apache.org>
>> <users@zeppelin.incubator.apache.org>
>> Subject:  R interpreters
>>
>> Hello:
>>
>> Looks like there are two R interpreters -
>> https://github.com/datalayer/zeppelin-R and
>> https://github.com/elbamos/Zeppelin-With-R
>>
>> From the docs, Zeppelin-With-R supports passing objects between languages
>> using registerTempTable.  Not sure if the other interpreter supports it or
>> what other differences are.  Has anyone looked into this?
>>
>>
>

Reply via email to