Github user maryannxue commented on a diff in the pull request:
https://github.com/apache/phoenix/pull/154#discussion_r57269833
--- Diff:
phoenix-core/src/main/java/org/apache/phoenix/compile/QueryCompiler.java ---
@@ -324,10 +328,13 @@ protected QueryPlan compileJoinQuery(StatementContext
context, List<Object> bind
QueryPlan plan = compileSingleFlatQuery(context, query, binds,
asSubquery, !asSubquery && joinTable.isAllLeftJoin(), null,
!table.isSubselect() && projectPKColumns ? tupleProjector : null, true);
Expression postJoinFilterExpression =
joinTable.compilePostFilterExpression(context, table);
Integer limit = null;
+ Integer offset = null;
if (!query.isAggregate() && !query.isDistinct() &&
query.getOrderBy().isEmpty()) {
limit = plan.getLimit();
+ offset = plan.getOffset();
}
- HashJoinInfo joinInfo = new HashJoinInfo(projectedTable,
joinIds, joinExpressions, joinTypes, starJoinVector, tables, fieldPositions,
postJoinFilterExpression, limit);
+ HashJoinInfo joinInfo = new HashJoinInfo(projectedTable,
joinIds, joinExpressions, joinTypes,
+ starJoinVector, tables, fieldPositions,
postJoinFilterExpression, limit, offset);
--- End diff --
The "limit" in HashJoinInfo is a little different than the original SQL
semantics. It specifies the max number of rows returned by a
HashJoinRegionScanner so that it doesn't do too much unnecessary work, and its
max rows per region, so in the end the client might get more rows than needed
and will do a final limit operation (all done by ScanPlan, don't worry). So,
here, I think the limit should be "offset + limit" and there is no need to add
an extra "offset" field to HashJoinInfo. Btw, change in
serialization/deserialization of HashJoinInfo would cause incompatibility
between different versions.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
