[
https://issues.apache.org/jira/browse/HIVE-27187?focusedWorklogId=854221&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-854221
]
ASF GitHub Bot logged work on HIVE-27187:
-----------------------------------------
Author: ASF GitHub Bot
Created on: 31/Mar/23 16:00
Start Date: 31/Mar/23 16:00
Worklog Time Spent: 10m
Work Description: deniskuzZ commented on code in PR #4166:
URL: https://github.com/apache/hive/pull/4166#discussion_r1154650166
##########
ql/src/java/org/apache/hadoop/hive/ql/ddl/view/materialized/alter/rebuild/AlterMaterializedViewRebuildAnalyzer.java:
##########
@@ -564,30 +582,38 @@ private void fixUpASTAggregateIncrementalRebuild(
HiveParser.TOK_INSERT, HiveParser.TOK_SELECT);
ASTNode selectExprNodeInputROJ = (ASTNode) ParseDriver.adaptor.create(
HiveParser.TOK_SELEXPR, "TOK_SELEXPR");
- ASTNode tableName = createRowIdNode(TableName.getDbTable(
+ astBuilder.createAcidSortNodes(TableName.getDbTable(
materializationNode.getChild(0).getText(),
- materializationNode.getChild(1).getText()));
- ParseDriver.adaptor.addChild(selectExprNodeInputROJ, tableName);
- ParseDriver.adaptor.addChild(selectNodeInputROJ, selectExprNodeInputROJ);
+ materializationNode.getChild(1).getText()))
+ .forEach(astNode ->
ParseDriver.adaptor.addChild(selectNodeInputROJ, astNode));
// 4) Transform first INSERT branch into an UPDATE
- // 4.1) Adding ROW__ID field
- ASTNode selectNodeInUpdate = (ASTNode) updateNode.getChild(1);
- if (selectNodeInUpdate.getType() != HiveParser.TOK_SELECT) {
- throw new SemanticException("TOK_SELECT expected in incremental
rewriting");
- }
- ASTNode selectExprNodeInUpdate = (ASTNode)
ParseDriver.adaptor.dupNode(selectExprNodeInputROJ);
- ParseDriver.adaptor.addChild(selectExprNodeInUpdate,
createRowIdNode((ASTNode) subqueryNodeInputROJ.getChild(1)));
- selectNodeInUpdate.insertChild(0, selectExprNodeInUpdate);
- // 4.2) Modifying filter condition.
+ // 4.1) Modifying filter condition.
ASTNode whereClauseInUpdate = findWhereClause(updateNode);
if (whereClauseInUpdate.getChild(0).getType() != HiveParser.KW_OR) {
throw new SemanticException("OR clause expected below TOK_WHERE in
incremental rewriting");
}
// We bypass the OR clause and select the first disjunct for the Update
branch
ParseDriver.adaptor.setChild(whereClauseInUpdate, 0,
disjuncts.get(Context.DestClausePrefix.UPDATE));
+ ASTNode updateDeleteNode = (ASTNode)
ParseDriver.adaptor.dupTree(updateNode);
+ // 4.2) Adding ROW__ID field
+ ASTNode selectNodeInUpdateDelete = (ASTNode) updateDeleteNode.getChild(1);
+ if (selectNodeInUpdateDelete.getType() != HiveParser.TOK_SELECT) {
+ throw new SemanticException("TOK_SELECT expected in incremental
rewriting");
+ }
+ // Remove children
+ while (selectNodeInUpdateDelete.getChildCount() > 0) {
+ selectNodeInUpdateDelete.deleteChild(0);
Review Comment:
can we remove all children in 1 go and don't call
freshenParentAndChildIndexes for every child?
Issue Time Tracking
-------------------
Worklog Id: (was: 854221)
Time Spent: 1.5h (was: 1h 20m)
> Incremental rebuild of materialized view having aggregate and stored by
> iceberg
> -------------------------------------------------------------------------------
>
> Key: HIVE-27187
> URL: https://issues.apache.org/jira/browse/HIVE-27187
> Project: Hive
> Issue Type: Improvement
> Components: Iceberg integration, Materialized views
> Reporter: Krisztian Kasa
> Assignee: Krisztian Kasa
> Priority: Major
> Labels: pull-request-available
> Time Spent: 1.5h
> Remaining Estimate: 0h
>
> Currently incremental rebuild of materialized view stored by iceberg which
> definition query contains aggregate operator is transformed to an insert
> overwrite statement which contains a union operator if the source tables
> contains insert operations only. One branch of the union scans the view the
> other produces the delta.
> This can be improved further: transform the statement to a multi insert
> statement representing a merge statement to insert new aggregations and
> update existing.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
