> > Now, I am not able to directly use my RDD object and have it implicitly > become a DataFrame. It can be used as a DataFrameHolder, of which I could > write: > > rdd.toDF.registerTempTable("foo") >
The rational here was that we added a lot of methods to DataFrame and made the implicits more powerful, but that increased the likelihood of accidental application of the implicit. I personally have had to explain the accidental application of implicits (and the confusing compiler messages that can result) to beginners so many times that we decided to remove the subtle conversion from RDD to DataFrame and instead make it explicit method call.