[SPARK-13948] MiMa check should catch if the visibility changes to private

MiMa excludes are currently generated using both the current Spark version's classes and Spark 1.2.0's classes, but this doesn't make sense: we should only be ignoring classes which were `private` in the previous Spark version, not classes which became private in the current version.

This patch updates `dev/mima` to only generate excludes with respect to the previous artifacts that MiMa checks against. It also updates `MimaBuild` so that `excludeClass` only applies directly to the class being excluded and not to its companion object (since a class and its companion object can have different accessibility).

Author: Josh Rosen <joshrosen@databricks.com>

Closes #11774 from JoshRosen/SPARK-13948.

1 files changed, 19 insertions, 0 deletions

diff --git a/project/MimaExcludes.scala b/project/MimaExcludes.scala
index 5e8d13b5b8..17ecb5f02a 100644
--- a/project/MimaExcludes.scala
+++ b/project/MimaExcludes.scala
@@ -224,6 +224,7 @@ object MimaExcludes {
         ProblemFilters.exclude[MissingClassProblem]("org.apache.spark.streaming.receiver.ActorReceiver$Supervisor")
       ) ++ Seq(
         // SPARK-12348 Remove deprecated Streaming APIs.
+        ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.streaming.dstream.DStream.foreach"),
         ProblemFilters.exclude[MissingMethodProblem]("org.apache.spark.streaming.StreamingContext.toPairDStreamFunctions"),
         ProblemFilters.exclude[MissingMethodProblem]("org.apache.spark.streaming.StreamingContext.toPairDStreamFunctions$default$4"),
         ProblemFilters.exclude[MissingMethodProblem]("org.apache.spark.streaming.StreamingContext.awaitTermination"),
@@ -540,6 +541,24 @@ object MimaExcludes {
         // SPARK-13927: add row/column iterator to local matrices
         ProblemFilters.exclude[MissingMethodProblem]("org.apache.spark.mllib.linalg.Matrix.rowIter"),
         ProblemFilters.exclude[MissingMethodProblem]("org.apache.spark.mllib.linalg.Matrix.colIter")
+      ) ++ Seq(
+        // SPARK-13948: MiMa Check should catch if the visibility change to `private`
+        // TODO(josh): Some of these may be legitimate incompatibilities; we should follow up before the 2.0.0 release
+        ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.sql.Dataset.toDS"),
+        ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.sql.sources.OutputWriterFactory.newInstance"),
+        ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.util.RpcUtils.askTimeout"),
+        ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.util.RpcUtils.lookupTimeout"),
+        ProblemFilters.exclude[IncompatibleMethTypeProblem]("org.apache.spark.ml.UnaryTransformer.transform"),
+        ProblemFilters.exclude[IncompatibleMethTypeProblem]("org.apache.spark.ml.classification.DecisionTreeClassifier.train"),
+        ProblemFilters.exclude[IncompatibleMethTypeProblem]("org.apache.spark.ml.classification.LogisticRegression.train"),
+        ProblemFilters.exclude[IncompatibleMethTypeProblem]("org.apache.spark.ml.regression.DecisionTreeRegressor.train"),
+        ProblemFilters.exclude[IncompatibleMethTypeProblem]("org.apache.spark.sql.Dataset.groupBy"),
+        ProblemFilters.exclude[IncompatibleResultTypeProblem]("org.apache.spark.sql.Dataset.groupBy"),
+        ProblemFilters.exclude[IncompatibleResultTypeProblem]("org.apache.spark.sql.Dataset.select"),
+        ProblemFilters.exclude[IncompatibleResultTypeProblem]("org.apache.spark.sql.Dataset.toDF"),
+        ProblemFilters.exclude[ReversedMissingMethodProblem]("org.apache.spark.Logging.initializeLogIfNecessary"),
+        ProblemFilters.exclude[ReversedMissingMethodProblem]("org.apache.spark.scheduler.SparkListenerEvent.logEvent"),
+        ProblemFilters.exclude[ReversedMissingMethodProblem]("org.apache.spark.sql.sources.OutputWriterFactory.newInstance")
       )
     case v if v.startsWith("1.6") =>
       Seq(