Github user xuchuanyin commented on a diff in the pull request:
https://github.com/apache/carbondata/pull/2609#discussion_r208112885
--- Diff:
integration/spark2/src/main/scala/org/apache/spark/sql/execution/strategy/DDLStrategy.scala
---
@@ -257,20 +256,13 @@ class DDLStrategy(sparkSession: SparkSession) extends
SparkStrategy {
if CarbonEnv.getInstance(sparkSession).carbonMetastore
.tableExists(tableName)(sparkSession) => {
- // TODO remove this limiation after streaming table support
'preaggregate' DataMap
- // if the table has 'preaggregate' DataMap, it doesn't support
streaming now
val carbonTable =
CarbonEnv.getInstance(sparkSession).carbonMetastore
.lookupRelation(tableName)(sparkSession).asInstanceOf[CarbonRelation].carbonTable
if (carbonTable != null &&
!carbonTable.getTableInfo.isTransactionalTable) {
throw new MalformedCarbonCommandException(
"Unsupported operation on non transactional table")
}
- if (carbonTable != null && !carbonTable.canAllow(carbonTable,
TableOperation.STREAMING)) {
--- End diff --
In the previous implementation, the `CarbonTable.canAllow` will only works
on the datamaps returned by `DataMapStorageManager.getInstance().getAllDataMap`
which are all index datamap.
Here the removed code means that there is no restriction for index map now,
so I think the modification do not affect the previous procedure.
As you mentioned about MV, maybe it is already a bug there.
---
