Github user liancheng commented on a diff in the pull request:
https://github.com/apache/spark/pull/4689#discussion_r24977332
--- Diff:
sql/core/src/test/scala/org/apache/spark/sql/CachedTableSuite.scala ---
@@ -280,4 +280,43 @@ class CachedTableSuite extends QueryTest {
assert(intercept[RuntimeException](table("t1")).getMessage.startsWith("Table
Not Found"))
assert(!isCached("t2"))
}
+
+ test("Drop cached temporary table when the table gets overwritten") {
+ val query1 = "SELECT key FROM testData LIMIT 10"
+ val df1 = sql(query1)
+ df1.registerTempTable("t1")
+ sql(s"CACHE TABLE t2 AS $query1")
+ assert(isCached("t2"))
+ // t1 will be cached too because it has the same plan as t2.
+ assert(isCached("t1"))
+ assert(cacheManager.lookupCachedData(df1).isDefined)
+
+ val query2 = "SELECT key FROM testData LIMIT 5"
+ val df2 = sql(query2)
+ sql(s"CACHE TABLE t2 AS $query2")
+ // t1 will not be cached because it has not been explicitly cached.
+ assert(!isCached("t1"))
+ assert(isCached("t2"))
+ assert(cacheManager.lookupCachedData(df2).isDefined)
+ dropTempTable("t2")
+
+ assert(cacheManager.lookupCachedData(df1).isEmpty)
+ assert(cacheManager.lookupCachedData(df2).isEmpty)
+ }
--- End diff --
How about this case:
```scala
val df1 = sql("SELECT * FROM testData LIMIT 10")
df1.registerTempTable("t1")
// Cache t1 explicitly
sql("CACHE TABLE t1")
// t1 and t2 share the same query plan
sql("CACHE TABLE t2 AS SELECT * FROM testData LIMIT 10")
// Replace t2 with a different query plan
sql("CACHE TABLE t2 AS SELECT * FROM testData LIMIT 5")
// Should t1 remain cached here?
```
To my understanding, with this PR, `t1` is implicitly uncached, which may
not be the behavior we want. I think we also need a map from query plans to
table names to prevent unexpected uncache operations.
---
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]
