有个朋友给了这样一段代码：

import org.apache.spark.sql.SparkSessionimport org.apache.spark.ml.linalg.{Vector, DenseVector}object HRDT {  def main(args : Array[String]) {    case class HRMeta(features:Vector, label:String)    val spark = SparkSession.builder()...    import spark.implicits._    val data = spark.sparkContext.textFile("...")        .map(l=>{          ...           new HRMeta(new DenseVector(Array(1.0,2.0)), "0")        }).toDF()  }}

编译的时候会报错：

[error] /data1/weibo_recmd/story/ctr_predict/code/HRDT.scala:18:12: value toDF is not a member of org.apache.spark.rdd.RDD[HRMeta][error] possible cause: maybe a semicolon is missing before `value toDF'?[error]         }).toDF()[error]            ^[error] one error found[error] (Compile / compileIncremental) Compilation failed[error] Total time: 12 s, completed Feb 9, 2018 1:50:13 PM

 

但如果把 case class 定义放到main方法外面，编译就没问题。

 

我猜测和DataSetHolder.toDF时候在scala里面对泛型的使用有关系，但没仔细调查验证。在此暂存一下