binning

[peller]

define discrete bins for charts

[peller]

should we do this in the kernel with dataframes or on the client with vega?

[dalogsdon]

I'd say kernel since the data could be relatively big, especially for Spark. We don't want to do any intensive computing in the client.

[peller]

@lbustelo

[peller]

so in addition to kernel-side binning, we will still need client-side handling to render properly, either setting up the same bins or transferring the information using discrete categories or passing the appropriate metadata to the charting engine