[
https://issues.apache.org/jira/browse/HIVE-26160?focusedWorklogId=762182&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-762182
]
ASF GitHub Bot logged work on HIVE-26160:
-----------------------------------------
Author: ASF GitHub Bot
Created on: 26/Apr/22 08:59
Start Date: 26/Apr/22 08:59
Worklog Time Spent: 10m
Work Description: kasakrisz commented on code in PR #3229:
URL: https://github.com/apache/hive/pull/3229#discussion_r858465153
##########
ql/src/java/org/apache/hadoop/hive/ql/parse/CalcitePlanner.java:
##########
@@ -2347,22 +2348,49 @@ private RelNode applyPostJoinOrderingTransform(RelNode
basePlan, RelMetadataProv
return basePlan;
}
+ /**
+ * Traverse the plan and collect table names from {@link TableScan}
operators
+ * Use this method if plan does not have any sub query.
+ * Use {@link CalcitePlannerAction#getAllTablesUsed(RelNode)} to include
sub-query expressions.
+ * @see HiveSubQueryRemoveRule
+ */
protected Set<TableName> getTablesUsed(RelNode plan) {
Set<TableName> tablesUsed = new HashSet<>();
new RelVisitor() {
@Override
public void visit(RelNode node, int ordinal, RelNode parent) {
- if (node instanceof TableScan) {
- TableScan ts = (TableScan) node;
- Table hiveTableMD = ((RelOptHiveTable)
ts.getTable()).getHiveTableMD();
- tablesUsed.add(hiveTableMD.getFullTableName());
- }
+ addUsedTable(node, tablesUsed);
super.visit(node, ordinal, parent);
}
}.go(plan);
return tablesUsed;
}
+ /**
+ * Traverse the plan including sub-query expressions and collect table
names from {@link TableScan} operators.
+ */
+ protected Set<TableName> getAllTablesUsed(RelNode plan) {
+ Set<TableName> tablesUsed = new HashSet<>();
+ new HiveSubQueryVisitor() {
Review Comment:
This seems to be a good idea. Let's see how it works: #3246
Issue Time Tracking
-------------------
Worklog Id: (was: 762182)
Time Spent: 40m (was: 0.5h)
> Materialized View rewrite does not check tables scanned in sub-query
> expressions
> --------------------------------------------------------------------------------
>
> Key: HIVE-26160
> URL: https://issues.apache.org/jira/browse/HIVE-26160
> Project: Hive
> Issue Type: Bug
> Components: CBO, Materialized views
> Reporter: Krisztian Kasa
> Assignee: Krisztian Kasa
> Priority: Major
> Labels: pull-request-available
> Time Spent: 40m
> Remaining Estimate: 0h
>
> Materialized View rewrite based on exact sql text match uses the initial CBO
> plan to explore possibilities to change the query plan or part of the plan to
> an MV scan.
> This algorithm requires the tables scanned by the original query plan. If the
> query contains sub query expressions the tables scanned by the sub query are
> not listed which can lead to rewrite the original plan to scan an outdated MV.
--
This message was sent by Atlassian Jira
(v8.20.7#820007)
