Experimental in Spark really just means that we are not promising binary compatibly for those functions in the 2.x release line. For Datasets in particular, we want a few releases to make sure the APIs don't have any major gaps before removing the experimental tag.
On Thu, Dec 15, 2016 at 1:17 PM, Ninad Shringarpure <ni...@cloudera.com> wrote: > Hi Team, > > When going through Dataset class for Spark 2.0 it comes across that both > overloaded map functions with encoder and without are marked as > experimental. > > Is there a reason and issues that developers whould be aware of when using > this for production applications. Also is there a "non-experimental" way of > using map function on Dataframe in Spark 2.0 > > Thanks, > Ninad >