AngersZhuuuu commented on a change in pull request #31378:
URL: https://github.com/apache/spark/pull/31378#discussion_r568352532
##########
File path:
sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/HiveCommandSuite.scala
##########
@@ -129,7 +129,7 @@ class HiveCommandSuite extends QueryTest with SQLTestUtils
with TestHiveSingleto
checkAnswer(
sql(s"SHOW TBLPROPERTIES parquet_tab1('my_key1')"),
- Row("v1") :: Nil
+ Row("my_key1", "v1") :: Nil
Review comment:
> > then shall we change the schema of the v2 command?
>
> IMO, change `ShowTablePropertiesCommand` is correct since hive's output
schema is `prpt_name` and `prpt_value`, I think hive is wrong before.
Also ping @wangyum @beliefer @viirya Also hope your suggestion since I
believe it's a hive's bug and we don't need to follow wrong behavior in hive
1.x and 2.x
##########
File path:
sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/HiveCommandSuite.scala
##########
@@ -129,7 +129,7 @@ class HiveCommandSuite extends QueryTest with SQLTestUtils
with TestHiveSingleto
checkAnswer(
sql(s"SHOW TBLPROPERTIES parquet_tab1('my_key1')"),
- Row("v1") :: Nil
+ Row("my_key1", "v1") :: Nil
Review comment:
> Changing the output schema looks fine to me. Could you describe this
change in the PR description?
Yea, updated, Do we need to change migration guide too?
##########
File path:
sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/HiveCommandSuite.scala
##########
@@ -129,7 +129,7 @@ class HiveCommandSuite extends QueryTest with SQLTestUtils
with TestHiveSingleto
checkAnswer(
sql(s"SHOW TBLPROPERTIES parquet_tab1('my_key1')"),
- Row("v1") :: Nil
+ Row("my_key1", "v1") :: Nil
Review comment:
> Yea let's add an item in the migration guide.
Done.
##########
File path: docs/sql-migration-guide.md
##########
@@ -56,6 +56,8 @@ license: |
In Spark 3.1 and earlier, table refreshing leaves dependents uncached.
- In Spark 3.2, the usage of `count(tblName.*)` is blocked to avoid
producing ambiguous results. Because `count(*)` and `count(tblName.*)` will
output differently if there is any null values. To restore the behavior before
Spark 3.2, you can set
`spark.sql.legacy.allowStarWithSingleTableIdentifierInCount` to `true`.
+
+ - In Spark 3.2, `ShowTablePropertiesCommand`'s output both show `key` and
`value` columns whether `propertyKey` is defined. In Spark 3.1 and earlier,
`ShowTablePropertiesCommand`'s output only show `value` column when
`propertyKey` is defined.
Review comment:
> `ShowTablePropertiesCommand` is an internal class name, so please use
"a `SHOW TBLPROPERTIES` clause` instead.
Done
##########
File path: docs/sql-migration-guide.md
##########
@@ -56,6 +56,8 @@ license: |
In Spark 3.1 and earlier, table refreshing leaves dependents uncached.
- In Spark 3.2, the usage of `count(tblName.*)` is blocked to avoid
producing ambiguous results. Because `count(*)` and `count(tblName.*)` will
output differently if there is any null values. To restore the behavior before
Spark 3.2, you can set
`spark.sql.legacy.allowStarWithSingleTableIdentifierInCount` to `true`.
+
+ - In Spark 3.2, a `SHOW TBLPROPERTIES` clause's output both show `key` and
`value` columns whether `propertyKey` is defined. In Spark 3.1 and earlier, a
`SHOW TBLPROPERTIES` clause's output only show a `value` column when
`propertyKey` is defined.
Review comment:
> `propertyKey` is also an internal variable name?
Yea ==
##########
File path: docs/sql-migration-guide.md
##########
@@ -56,6 +56,8 @@ license: |
In Spark 3.1 and earlier, table refreshing leaves dependents uncached.
- In Spark 3.2, the usage of `count(tblName.*)` is blocked to avoid
producing ambiguous results. Because `count(*)` and `count(tblName.*)` will
output differently if there is any null values. To restore the behavior before
Spark 3.2, you can set
`spark.sql.legacy.allowStarWithSingleTableIdentifierInCount` to `true`.
+
+ - In Spark 3.2, a `SHOW TBLPROPERTIES` clause's output both show `key` and
`value` columns whether `propertyKey` is defined. In Spark 3.1 and earlier, a
`SHOW TBLPROPERTIES` clause's output only show a `value` column when
`propertyKey` is defined.
Review comment:
> "output both show `key` and `value` columns" => "output shows `key`
and `value` columns"
Done
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
