CodingCat commented on code in PR #5083:
URL: https://github.com/apache/iceberg/pull/5083#discussion_r908675283
##########
spark/v3.2/spark-extensions/src/main/scala/org/apache/spark/sql/catalyst/analysis/RewriteDeleteFromTable.scala:
##########
@@ -49,25 +51,42 @@ import
org.apache.spark.sql.execution.datasources.v2.DataSourceV2Relation
object RewriteDeleteFromTable extends RewriteRowLevelCommand {
override def apply(plan: LogicalPlan): LogicalPlan = plan resolveOperators {
- case d @ DeleteFromIcebergTable(aliasedTable, Some(cond), None) if
d.resolved =>
+ case d @ DeleteFromIcebergTable(aliasedTable, Some(_), None) if d.resolved
=>
EliminateSubqueryAliases(aliasedTable) match {
case r @ DataSourceV2Relation(tbl: SupportsRowLevelOperations, _, _,
_, _) =>
- val operation = buildRowLevelOperation(tbl, DELETE)
- val table = RowLevelOperationTable(tbl, operation)
- val rewritePlan = operation match {
- case _: SupportsDelta =>
- buildWriteDeltaPlan(r, table, cond)
- case _ =>
- buildReplaceDataPlan(r, table, cond)
+ rewriteDeleteFromTable(d, r, tbl)
+ case v: View =>
+ val viewChild = v.child
+ val icebergTableView = v.children.size == 1 &&
viewChild.isInstanceOf[DataSourceV2Relation] &&
Review Comment:
the major reason to use view is that we have no ways to refer to a table
created with HadoopTables.create()..which is usually a pure directory path....
so we have to load the table and then register it as a view, and then we can
refer it with the view name in the statement like MERGE INTO, etc.
--
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]
