search

EclairJS / eclairjs-nashorn

JavaScript API for Apache Spark
Apache License 2.0
94 stars 11 forks source link

Investigate JavaScript support for Jupyter notebook without Spark #228

Open david-f opened 8 years ago

david-f commented 8 years ago

Jupyter notebooks most often run with a simple Python or R interpreter and without Spark as a big compute engine. What would it take to enable EclairJS to run in such an environment?

billreed63 commented 8 years ago

I am not a Jupyter notebooks expert (maybe @bpburns can correct me were I go wrong), the way I understand Jupyter notebooks is in order to support other languages is to add iPython/Jupyter kernels. To run R for example the user would need to install the the iRKernel for Jupyter. This would enable R code in the user's notebook, this would not allow them to use SparkR from the notebook.

EclairJS-nashorn support Jupyter notebooks through the Apache Toree Jupyter kernel. This allows the user to write EclairJS-nashorn code that will run on the Spark cluster.

If the desire is to run JavaScript in Jupyter notebooks without Spark then the user could install iJavaScript

billreed63 commented 8 years ago

We have a EclairJS kernel for Jupyter notebooks, but "magics" are not working. We need to get magics working do that we support HTML and D3 for visualization of data.

billreed63 commented 8 years ago

to integrate with Jupyter notebooks. EclairJS extends Toree's org.apache.toree.interpreter.Interpreter for EclairJS support in Jupyter notebooks. Currently Apache Toree only support "magics" in the scala interpreter, in order for EclairJS notebooks to use "magics" we would need to develop a feature enhancement for Apache Toree to handle "magics" globally. Apache Toree has offered to provide guidance but not do the actual development. If EclairJS want to put a Pull Request together they would merge it.