Use '1-x' value of bulkLoadPortion x

szarnyasg · szarnyasg · commit d543f78b5da8 · 2021-07-13T00:36:46.000+02:00
diff --git a/src/main/scala/ldbc/snb/datagen/spark/LdbcDatagen.scala b/src/main/scala/ldbc/snb/datagen/spark/LdbcDatagen.scala
@@ -17,7 +17,7 @@ object LdbcDatagen extends SparkApp {
     params: Map[String, String] = Map.empty,
     paramFile: Option[String] = None,
     outputDir: String = "out",
-    bulkloadPortion: Double = 0.1,
+    bulkloadPortion: Double = 0.9,
     explodeEdges: Boolean = false,
     explodeAttrs: Boolean = false,
     keepImplicitDeletes: Boolean = false,
diff --git a/src/main/scala/ldbc/snb/datagen/transformation/transform/Interactive.scala b/src/main/scala/ldbc/snb/datagen/transformation/transform/Interactive.scala
@@ -21,7 +21,7 @@ private object Interactive {
   }
 
   def calculateBulkLoadThreshold(bulkLoadPortion: Double, simulationStart: Long, simulationEnd: Long) = {
-    (simulationEnd - ((simulationEnd - simulationStart) * bulkLoadPortion).toLong)
+    (simulationEnd - ((simulationEnd - simulationStart) * (1 - bulkLoadPortion)).toLong)
   }
 
   def snapshotPart(tpe: EntityType, df: DataFrame, bulkLoadThreshold: Long, filterDeletion: Boolean) = {

Original file line number	Diff line number	Diff line change
`@@ -21,7 +21,7 @@ private object Interactive {`
`21`	`21`	`}`
`22`	`22`
`23`	`23`	`def calculateBulkLoadThreshold(bulkLoadPortion: Double, simulationStart: Long, simulationEnd: Long) = {`
`24`		`- (simulationEnd - ((simulationEnd - simulationStart) * bulkLoadPortion).toLong)`
	`24`	`+ (simulationEnd - ((simulationEnd - simulationStart) * (1 - bulkLoadPortion)).toLong)`
`25`	`25`	`}`
`26`	`26`
`27`	`27`	`def snapshotPart(tpe: EntityType, df: DataFrame, bulkLoadThreshold: Long, filterDeletion: Boolean) = {`