Github user rdblue commented on a diff in the pull request:
    --- Diff: 
    @@ -23,11 +23,11 @@ import org.apache.spark.sql.execution.SparkPlan
     object DataSourceV2Strategy extends Strategy {
       override def apply(plan: LogicalPlan): Seq[SparkPlan] = plan match {
    -    case relation: DataSourceV2Relation =>
    -      DataSourceV2ScanExec(relation.output, relation.reader) :: Nil
    +    case r: DataSourceV2Relation =>
    --- End diff --
    I'm pointing it out because I think it is significant. This is also why I 
sent a note to the dev list. Changes like this make it much harder to work with 
Spark because little things change that are not necessary and cause conflicts. 
That's a problem not just when I'm maintaining a branch, but also when I'm 
trying to get a PR committed. These changes cause more work because we have to 
rebase PRs and update for variables that have changed names in the name of 
    In this particular case, I would probably wrap the line that is too long. 
If you want to rename to fix it, I'd consider that reasonable. But all the 
other cases that aren't necessary should be reverted.


To unsubscribe, e-mail:
For additional commands, e-mail:

Reply via email to