Github user hvanhovell commented on a diff in the pull request:
https://github.com/apache/spark/pull/10605#discussion_r49219114
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/Window.scala ---
@@ -307,27 +314,63 @@ case class Window(
// Collect all the rows in the current partition.
// Before we start to fetch new input rows, make a copy of
nextGroup.
val currentGroup = nextGroup.copy()
- rows.clear()
+
+ // clear last partition
+ if (sorter != null) {
+ // the last sorter of this task will be cleaned up via task
completion listener
+ sorter.cleanupResources()
+ sorter = null
+ } else {
+ rows.clear()
+ }
+
while (nextRowAvailable && nextGroup == currentGroup) {
- rows += nextRow.copy()
+ if (sorter == null) {
+ rows += nextRow.copy()
+
+ if (rows.length >= 4096) {
+ // We will not sort the rows, so prefixComparator and
recordComparator are null.
+ sorter = UnsafeExternalSorter.create(
+ TaskContext.get().taskMemoryManager(),
+ SparkEnv.get.blockManager,
+ TaskContext.get(),
+ null,
+ null,
+ 1024,
+ SparkEnv.get.memoryManager.pageSizeBytes)
+ rows.foreach { r =>
+ sorter.insertRecord(r.getBaseObject, r.getBaseOffset,
r.getSizeInBytes, 0)
+ }
+ rows.clear()
+ }
+ } else {
+ sorter.insertRecord(nextRow.getBaseObject,
nextRow.getBaseOffset,
+ nextRow.getSizeInBytes, 0)
+ }
fetchNextRow()
}
+ if (sorter != null) {
+ rowBuffer = new ExternalRowBuffer(sorter, inputFields)
+ } else {
+ rowBuffer = new ArrayRowBuffer(rows)
+ }
// Setup the frames.
var i = 0
while (i < numFrames) {
- frames(i).prepare(rows)
+ frames(i).prepare(rowBuffer.copy())
--- End diff --
@tedyu both buffers implementations are statefull (they contain an
iterator). Each frame traverses the partition in a different way, so we cannot
share iterators between them.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
