This is the reverse of the task in SPARKR-187 where we take an existing RRRD and apply a schema to convert it to a DataFrame
This change should include support for specifying a schema using common R types such as "character", "int" etc.
Davies Could you co-ordinate with Chris Freeman to make sure there isn't any duplicated work ?
Also I wrote up a couple of gists sketching the implementation at https://gist.github.com/shivaram/92c8f3e6431e647912f3 and https://gist.github.com/shivaram/ac1b35b18aa55c3717e0 that may be useful
Fixed by https://github.com/amplab-extras/SparkR-pkg/pull/212