[
https://issues.apache.org/jira/browse/CASSANDRA-11310?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15197656#comment-15197656
]
Benjamin Lerer commented on CASSANDRA-11310:
--------------------------------------------
While reviewing your latest patch I realized that it is important to make sure
that {{usesSecondaryIndexing}} is properly set as it influences the way
{{SelectStatement}} will process the queries. It seems that it is already wrong
today. If a query try to perform filtering with a {{CONTAINS}} on a non-primary
key column, {{SelectStatement}} will believe that an index is used even if it
is not the case. The query will be executed but not necessarily in the expected
way and the order by clause will be rejected :-(.
To make the code consistent I believe that {{StatementRestrictions}} should
have an {{useFiltering}} boolean variable, similar to
{{usesSecondaryIndexing}}, to keep tract of the fact that the query will
perform filtering. It might be nice to expose it and have a test in
{{SelectStatement::getRowFilter}} to return an empty {{RowFilter}} if neither
filtering nor secondary indices are used.
I do not think that the {{allowFiltering}} variable need to be stored as a
field in {{StatementRestrictions}}. It can be passed as a variable.
On the testing side, I think it would be good to:
* run your tests on {{COMPACT}} tables as well
* test with {{ORDER BY}} clauses for single and multi partition queries
* test with filtering on clustering columns and non-primary key columns /
static columns
Problems tend to show up when several functionnalities are used together, so it
is best to test the combinations which seems a bit risky. :-)
> Allow filtering on clustering columns for queries without secondary indexes
> ---------------------------------------------------------------------------
>
> Key: CASSANDRA-11310
> URL: https://issues.apache.org/jira/browse/CASSANDRA-11310
> Project: Cassandra
> Issue Type: Improvement
> Components: CQL
> Reporter: Benjamin Lerer
> Assignee: Alex Petrov
> Labels: doc-impacting
> Fix For: 3.x
>
>
> Since CASSANDRA-6377 queries without index filtering non-primary key columns
> are fully supported.
> It makes sense to also support filtering on clustering-columns.
> {code}
> CREATE TABLE emp_table2 (
> empID int,
> firstname text,
> lastname text,
> b_mon text,
> b_day text,
> b_yr text,
> PRIMARY KEY (empID, b_yr, b_mon, b_day));
> SELECT b_mon,b_day,b_yr,firstname,lastname FROM emp_table2
> WHERE b_mon='oct' ALLOW FILTERING;
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
