[SPARK-26029][BUILD][2.4] Bump previousSparkVersion in MimaBuild.scala to be 2.3.0

## What changes were proposed in this pull request?

Although it's a little late, we should still update mima for branch 2.4, to avoid future breaking changes.

Note that, when merging, we should forward port it to master branch, so that the excluding rules are still in `v24excludes`.

TODO: update the release process document to mention about mima update.

## How was this patch tested?

N/A

Closes apache#23015 from cloud-fan/mima-2.4.

Authored-by: Wenchen Fan <[email protected]>
Signed-off-by: Wenchen Fan <[email protected]>

Author: cloud-fan

project/MimaBuild.scala

@@ -88,7 +88,7 @@ object MimaBuild {
 
   def mimaSettings(sparkHome: File, projectRef: ProjectRef) = {
     val organization = "org.apache.spark"
-    val previousSparkVersion = "2.2.0"
+    val previousSparkVersion = "2.3.0"
     val project = projectRef.project
     val fullId = "spark-" + project + "_2.11"
     mimaDefaultSettings ++

project/MimaExcludes.scala

@@ -105,7 +105,50 @@ object MimaExcludes {
     ProblemFilters.exclude[InheritedNewAbstractMethodProblem]("org.apache.spark.ml.param.shared.HasValidationIndicatorCol.validationIndicatorCol"),
 
     // [SPARK-23042] Use OneHotEncoderModel to encode labels in MultilayerPerceptronClassifier
-    ProblemFilters.exclude[MissingClassProblem]("org.apache.spark.ml.classification.LabelConverter")
+    ProblemFilters.exclude[MissingClassProblem]("org.apache.spark.ml.classification.LabelConverter"),
+
+    // [SPARK-21842][MESOS] Support Kerberos ticket renewal and creation in Mesos
+    ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.deploy.SparkHadoopUtil.getDateOfNextUpdate"),
+
+    // [SPARK-23366] Improve hot reading path in ReadAheadInputStream
+    ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.io.ReadAheadInputStream.this"),
+
+    // [SPARK-22941][CORE] Do not exit JVM when submit fails with in-process launcher.
+    ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.deploy.SparkSubmit.addJarToClasspath"),
+    ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.deploy.SparkSubmit.mergeFileLists"),
+    ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.deploy.SparkSubmit.prepareSubmitEnvironment$default$2"),
+
+    // Data Source V2 API changes
+    // TODO: they are unstable APIs and should not be tracked by mima.
+    ProblemFilters.exclude[MissingClassProblem]("org.apache.spark.sql.sources.v2.ReadSupportWithSchema"),
+    ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.sql.sources.v2.reader.SupportsScanColumnarBatch.createDataReaderFactories"),
+    ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.sql.sources.v2.reader.SupportsScanColumnarBatch.createBatchDataReaderFactories"),
+    ProblemFilters.exclude[ReversedMissingMethodProblem]("org.apache.spark.sql.sources.v2.reader.SupportsScanColumnarBatch.planBatchInputPartitions"),
+    ProblemFilters.exclude[MissingClassProblem]("org.apache.spark.sql.sources.v2.reader.SupportsScanUnsafeRow"),
+    ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.sql.sources.v2.reader.DataSourceReader.createDataReaderFactories"),
+    ProblemFilters.exclude[ReversedMissingMethodProblem]("org.apache.spark.sql.sources.v2.reader.DataSourceReader.planInputPartitions"),
+    ProblemFilters.exclude[MissingClassProblem]("org.apache.spark.sql.sources.v2.reader.SupportsPushDownCatalystFilters"),
+    ProblemFilters.exclude[MissingClassProblem]("org.apache.spark.sql.sources.v2.reader.DataReader"),
+    ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.sql.sources.v2.reader.SupportsReportStatistics.getStatistics"),
+    ProblemFilters.exclude[ReversedMissingMethodProblem]("org.apache.spark.sql.sources.v2.reader.SupportsReportStatistics.estimateStatistics"),
+    ProblemFilters.exclude[MissingClassProblem]("org.apache.spark.sql.sources.v2.reader.DataReaderFactory"),
+    ProblemFilters.exclude[MissingClassProblem]("org.apache.spark.sql.sources.v2.reader.streaming.ContinuousDataReader"),
+    ProblemFilters.exclude[MissingClassProblem]("org.apache.spark.sql.sources.v2.writer.SupportsWriteInternalRow"),
+    ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.sql.sources.v2.writer.DataWriterFactory.createDataWriter"),
+    ProblemFilters.exclude[ReversedMissingMethodProblem]("org.apache.spark.sql.sources.v2.writer.DataWriterFactory.createDataWriter"),
+
+    // Changes to HasRawPredictionCol.
+    ProblemFilters.exclude[InheritedNewAbstractMethodProblem]("org.apache.spark.ml.param.shared.HasRawPredictionCol.rawPredictionCol"),
+    ProblemFilters.exclude[InheritedNewAbstractMethodProblem]("org.apache.spark.ml.param.shared.HasRawPredictionCol.org$apache$spark$ml$param$shared$HasRawPredictionCol$_setter_$rawPredictionCol_="),
+    ProblemFilters.exclude[InheritedNewAbstractMethodProblem]("org.apache.spark.ml.param.shared.HasRawPredictionCol.getRawPredictionCol"),
+
+    // [SPARK-15526][ML][FOLLOWUP] Make JPMML provided scope to avoid including unshaded JARs
+    (problem: Problem) => problem match {
+      case MissingClassProblem(cls) =>
+        !cls.fullName.startsWith("org.spark_project.jpmml") &&
+          !cls.fullName.startsWith("org.spark_project.dmg.pmml")
+      case _ => true
+    }
   )
 
   // Exclude rules for 2.3.x