Github user cloud-fan commented on a diff in the pull request:
https://github.com/apache/spark/pull/16933#discussion_r104786851
--- Diff:
sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/optimizer/CollapseRepartitionSuite.scala
---
@@ -32,47 +32,180 @@ class CollapseRepartitionSuite extends PlanTest {
val testRelation = LocalRelation('a.int, 'b.int)
+
+ test("collapse two adjacent coalesces into one") {
+ // Always respects the top coalesces amd removes useless coalesce
below coalesce
+ val query1 = testRelation
+ .coalesce(10)
+ .coalesce(20)
+ val query2 = testRelation
+ .coalesce(30)
+ .coalesce(20)
+
+ val optimized1 = Optimize.execute(query1.analyze)
+ val optimized2 = Optimize.execute(query2.analyze)
+
+ val correctAnswer = testRelation.coalesce(20).analyze
+
+ comparePlans(optimized1, correctAnswer)
+ comparePlans(optimized2, correctAnswer)
+ }
+
test("collapse two adjacent repartitions into one") {
- val query = testRelation
+ // Always respects the top repartition amd removes useless repartition
below repartition
+ val query1 = testRelation
+ .repartition(10)
+ .repartition(20)
+ val query2 = testRelation
+ .repartition(30)
+ .repartition(20)
+
+ val optimized1 = Optimize.execute(query1.analyze)
+ val optimized2 = Optimize.execute(query2.analyze)
+ val correctAnswer = testRelation.repartition(20).analyze
+
+ comparePlans(optimized1, correctAnswer)
+ comparePlans(optimized2, correctAnswer)
+ }
+
+ test("coalesce above repartition") {
+ // Remove useless coalesce above repartition
+ val query1 = testRelation
.repartition(10)
+ .coalesce(20)
+
+ val optimized1 = Optimize.execute(query1.analyze)
+ val correctAnswer1 = testRelation.repartition(10).analyze
+
+ comparePlans(optimized1, correctAnswer1)
+
+ // No change in this case
+ val query2 = testRelation
+ .repartition(30)
+ .coalesce(20)
+
+ val optimized2 = Optimize.execute(query2.analyze)
+ val correctAnswer2 = query2.analyze
+
+ comparePlans(optimized2, correctAnswer2)
+ }
+
+ test("repartition above coalesce") {
+ // Always respects the top repartition amd removes useless coalesce
below repartition
+ val query1 = testRelation
+ .coalesce(10)
+ .repartition(20)
+ // Remove useless coalesce above repartition
+ val query2 = testRelation
+ .coalesce(30)
.repartition(20)
- val optimized = Optimize.execute(query.analyze)
+ val optimized1 = Optimize.execute(query1.analyze)
+ val optimized2 = Optimize.execute(query2.analyze)
+
val correctAnswer = testRelation.repartition(20).analyze
- comparePlans(optimized, correctAnswer)
+ comparePlans(optimized1, correctAnswer)
+ comparePlans(optimized2, correctAnswer)
}
- test("collapse repartition and repartitionBy into one") {
- val query = testRelation
+ test("repartitionBy above repartition") {
+ val query1 = testRelation
.repartition(10)
.distribute('a)(20)
- val optimized = Optimize.execute(query.analyze)
- val correctAnswer = testRelation.distribute('a)(20).analyze
+ val optimized1 = Optimize.execute(query1.analyze)
+ val correctAnswer1 = testRelation.distribute('a)(20).analyze
+
+ comparePlans(optimized1, correctAnswer1)
+
+ val query2 = testRelation
+ .repartition(30)
+ .distribute('a)(20)
+
+ val optimized2 = Optimize.execute(query2.analyze)
+ val correctAnswer2 = testRelation.distribute('a)(20).analyze
- comparePlans(optimized, correctAnswer)
+ comparePlans(optimized2, correctAnswer2)
}
- test("collapse repartitionBy and repartition into one") {
- val query = testRelation
+ test("repartitionBy above coalesce") {
+ val query1 = testRelation
+ .coalesce(10)
+ .distribute('a)(20)
+
+ val optimized1 = Optimize.execute(query1.analyze)
+ val correctAnswer1 = testRelation.distribute('a)(20).analyze
+
+ comparePlans(optimized1, correctAnswer1)
+
+ val query2 = testRelation
+ .coalesce(20)
+ .distribute('a)(30)
+
+ val optimized2 = Optimize.execute(query2.analyze)
+ val correctAnswer2 = testRelation.distribute('a)(30).analyze
+
+ comparePlans(optimized2, correctAnswer2)
+ }
+
+ test("repartition above repartitionBy") {
+ val query1 = testRelation
.distribute('a)(20)
.repartition(10)
- val optimized = Optimize.execute(query.analyze)
- val correctAnswer = testRelation.distribute('a)(10).analyze
+ val optimized1 = Optimize.execute(query1.analyze)
+ val correctAnswer1 = testRelation.distribute('a)(10).analyze
--- End diff --
My concern is, optimization should not change the result.
`relation.distributeBy('a, 10).repartition(10)` should have same result of
`relation.repartition(10)`, instead of `relation.distributeBy('a, 10)`. It's
not about which one is cheaper, we should not surprise users.
---
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]
