[jira] [Created] (CALCITE-3642) Update cassandra tests upgrade from junit4 to junit5
Forward Xu created CALCITE-3642: --- Summary: Update cassandra tests upgrade from junit4 to junit5 Key: CALCITE-3642 URL: https://issues.apache.org/jira/browse/CALCITE-3642 Project: Calcite Issue Type: Improvement Reporter: Forward Xu Update `Cassandra` tests upgrade from junit4 to junit5 -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Created] (CALCITE-3641) Oracle XMLCOMMENT Function Support
Ritesh created CALCITE-3641: --- Summary: Oracle XMLCOMMENT Function Support Key: CALCITE-3641 URL: https://issues.apache.org/jira/browse/CALCITE-3641 Project: Calcite Issue Type: Sub-task Reporter: Ritesh Assignee: Ritesh [https://docs.oracle.com/cd/B19306_01/server.102/b14200/functions218.htm] -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Created] (CALCITE-3640) Oracle EXISTSNODE Function Support
Ritesh created CALCITE-3640: --- Summary: Oracle EXISTSNODE Function Support Key: CALCITE-3640 URL: https://issues.apache.org/jira/browse/CALCITE-3640 Project: Calcite Issue Type: Sub-task Reporter: Ritesh [https://docs.oracle.com/cd/B19306_01/server.102/b14200/functions048.htm] -- This message was sent by Atlassian Jira (v8.3.4#803005)
Re: License header vs package-info.java vs whitespace
Thanks for updating! It is useful to know that the empty line has a usage. -Rui On Fri, Dec 27, 2019 at 12:30 PM Vladimir Sitnikov < sitnikov.vladi...@gmail.com> wrote: > Ok, it looks like I found yet another reasoning for having a blank line > between consequent comments. > It simplifies shell-like comment processing. > > For instance, the following is non-trivial for an automatic header update > (it is not clear where copyright ends and the next comment starts): > > # > # Licensed under Apache 2.0 > # > # Root logger is configured at INFO and is sent to A1 > log4j.rootLogger=INFO, A1 > > However, if we add a blank line in-between the two comments, then it > becomes much easier for machine processing: > > # > # Licensed under Apache 2.0 > # > > # Root logger is configured at INFO and is sent to A1 > log4j.rootLogger=INFO, A1 > > Vladimir >
Re: License header vs package-info.java vs whitespace
Ok, it looks like I found yet another reasoning for having a blank line between consequent comments. It simplifies shell-like comment processing. For instance, the following is non-trivial for an automatic header update (it is not clear where copyright ends and the next comment starts): # # Licensed under Apache 2.0 # # Root logger is configured at INFO and is sent to A1 log4j.rootLogger=INFO, A1 However, if we add a blank line in-between the two comments, then it becomes much easier for machine processing: # # Licensed under Apache 2.0 # # Root logger is configured at INFO and is sent to A1 log4j.rootLogger=INFO, A1 Vladimir
[jira] [Created] (CALCITE-3639) JoinConditionPushRule fail to push filter to inputs
Wang Yanlin created CALCITE-3639:
Summary: JoinConditionPushRule fail to push filter to inputs
Key: CALCITE-3639
URL: https://issues.apache.org/jira/browse/CALCITE-3639
Project: Calcite
Issue Type: Bug
Reporter: Wang Yanlin
The relnode(without applying optimize rules) for the sql
{code:java}
String sql = "select empno, emp.deptno from emp left join dept\n"
+ " on emp.deptno = dept.deptno and empno = 10 and dept.deptno = 20";
{code}
is
{code:java}
LogicalProject(EMPNO=[$0], DEPTNO=[$7])
LogicalJoin(condition=[AND(=($7, $9), =($0, 10), =($9, 20))], joinType=[left])
LogicalTableScan(table=[[CATALOG, SALES, EMP]])
LogicalTableScan(table=[[CATALOG, SALES, DEPT]])
{code}
After optimized with *JoinConditionPushRule*, the relnode becomes
{code:java}
LogicalProject(EMPNO=[$0], DEPTNO=[$7])
LogicalJoin(condition=[AND(=($7, $9), =($0, 10))], joinType=[left])
LogicalTableScan(table=[[CATALOG, SALES, EMP]])
LogicalFilter(condition=[=($0, 20)])
LogicalTableScan(table=[[CATALOG, SALES, DEPT]])
{code}
The optimize rule failed to push *empno = 10* to the left input, the better
relnode should be
{code:java}
LogicalProject(EMPNO=[$0], DEPTNO=[$7])
LogicalJoin(condition=[AND(=($7, $9))], joinType=[left])
LogicalFilter(condition=[=($0, 10)])
LogicalTableScan(table=[[CATALOG, SALES, EMP]])
LogicalFilter(condition=[=($0, 20)])
LogicalTableScan(table=[[CATALOG, SALES, DEPT]])
{code}
Add this test case to reproduce
{code:java}
// RelOptRulesTest
@Test public void testFilterInLeftJoin() {
String sql = "select empno, emp.deptno from emp left join dept\n"
+ " on emp.deptno = dept.deptno and empno = 10 and dept.deptno = 20";
sql(sql).withRule(FilterJoinRule.JOIN).check();
}
// data for this case
{code}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
