jingz-db commented on code in PR #47107:
URL: https://github.com/apache/spark/pull/47107#discussion_r1667125868
##########
sql/core/src/main/scala/org/apache/spark/sql/execution/streaming/state/RocksDBStateStoreProvider.scala:
##########
@@ -57,20 +60,22 @@ private[sql] class RocksDBStateStoreProvider
keyStateEncoderSpec: KeyStateEncoderSpec,
useMultipleValuesPerKey: Boolean = false,
isInternal: Boolean = false): Unit = {
- verify(colFamilyName != StateStore.DEFAULT_COL_FAMILY_NAME,
- s"Failed to create column family with reserved_name=$colFamilyName")
- verify(useColumnFamilies, "Column families are not supported in this
store")
- rocksDB.createColFamilyIfAbsent(colFamilyName, isInternal)
+ ColumnFamilyUtils.createColFamilyIfAbsent(colFamilyName, isInternal)
+
keyValueEncoderMap.putIfAbsent(colFamilyName,
- (RocksDBStateEncoder.getKeyEncoder(keyStateEncoderSpec),
+ (RocksDBStateEncoder.getKeyEncoder(keyStateEncoderSpec,
useColumnFamilies),
Review Comment:
We won't have case where we assign different IDs to the same CF during
lifecycle of the encoder instance. In my new refactored RocksDBKeyEncoder
class, we'll pass in a virtual column family id during column family creation
time. This way we can get rid of looking up column family id every time in the
`RocksDBStateStoreProvider`.
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
