[jira] [Commented] (PHOENIX-3382) Query using Row Value Constructor comprised of non leading PK columns returns incorrect results
[ https://issues.apache.org/jira/browse/PHOENIX-3382?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15607068#comment-15607068 ] Hudson commented on PHOENIX-3382: - FAILURE: Integrated in Jenkins build Phoenix-4.8-HBase-1.2 #41 (See [https://builds.apache.org/job/Phoenix-4.8-HBase-1.2/41/]) PHOENIX-3382 Query using Row Value Constructor comprised of non leading (jamestaylor: rev b52321d7769ea785a4336c36242497a47be711d3) * (edit) phoenix-core/src/main/java/org/apache/phoenix/compile/WhereOptimizer.java * (edit) phoenix-core/src/it/java/org/apache/phoenix/end2end/QueryMoreIT.java * (edit) phoenix-core/src/test/java/org/apache/phoenix/compile/WhereOptimizerTest.java > Query using Row Value Constructor comprised of non leading PK columns returns > incorrect results > --- > > Key: PHOENIX-3382 > URL: https://issues.apache.org/jira/browse/PHOENIX-3382 > Project: Phoenix > Issue Type: Bug >Affects Versions: 4.8.0 >Reporter: Joel Palmert >Assignee: James Taylor >Priority: Critical > Fix For: 4.9.0, 4.8.2 > > Attachments: PHOENIX-3382.patch, PHOENIX-3382_wip.patch > > > I'm trying to use Row Value Constructors to do paging in a query in the below > table: > # TABLE # > CREATE TABLE IF NOT EXISTS COMMUNITIES.TEST ( > ORGANIZATION_ID CHAR(15) NOT NULL, > SCORE DOUBLE NOT NULL, > ENTITY_ID CHAR(15) NOT NULL > CONSTRAINT TEST_PK PRIMARY KEY ( > ORGANIZATION_ID, > SCORE DESC, > ENTITY_ID DESC > ) > ) VERSIONS=1, MULTI_TENANT=TRUE, REPLICATION_SCOPE=1, TTL=43200 > UPSERT INTO COMMUNITIES.TEST VALUES ('org1',3,'01'); > UPSERT INTO COMMUNITIES.TEST VALUES ('org1',2,'04'); > UPSERT INTO COMMUNITIES.TEST VALUES ('org1',2,'03'); > UPSERT INTO COMMUNITIES.TEST VALUES ('org1',1,'02'); > # QUERY # > SELECT entity_id, score > FROM communities.test > WHERE organization_id = 'org1' > AND (score, entity_id) < (2, '04') > ORDER BY score DESC, entity_id DESC > LIMIT 3 > I'm specifying the score and entity_id of the second row so I'm expecting it > to return the 2 last values (with id 03 and 02). However, it returns the 3 > first values (and if it wasn't for the LIMIT 3 it would just return all). -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (PHOENIX-3382) Query using Row Value Constructor comprised of non leading PK columns returns incorrect results
[ https://issues.apache.org/jira/browse/PHOENIX-3382?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15579204#comment-15579204 ] Hudson commented on PHOENIX-3382: - SUCCESS: Integrated in Jenkins build Phoenix-master #1442 (See [https://builds.apache.org/job/Phoenix-master/1442/]) PHOENIX-3382 Query using Row Value Constructor comprised of non leading (jamestaylor: rev 3e4aec1bc440ad503d5058237e3b508392df9266) * (edit) phoenix-core/src/test/java/org/apache/phoenix/compile/WhereOptimizerTest.java * (edit) phoenix-core/src/main/java/org/apache/phoenix/compile/WhereOptimizer.java * (edit) phoenix-core/src/it/java/org/apache/phoenix/end2end/QueryMoreIT.java > Query using Row Value Constructor comprised of non leading PK columns returns > incorrect results > --- > > Key: PHOENIX-3382 > URL: https://issues.apache.org/jira/browse/PHOENIX-3382 > Project: Phoenix > Issue Type: Bug >Affects Versions: 4.8.0 >Reporter: Joel Palmert >Priority: Critical > Attachments: PHOENIX-3382.patch, PHOENIX-3382_wip.patch > > > I'm trying to use Row Value Constructors to do paging in a query in the below > table: > # TABLE # > CREATE TABLE IF NOT EXISTS COMMUNITIES.TEST ( > ORGANIZATION_ID CHAR(15) NOT NULL, > SCORE DOUBLE NOT NULL, > ENTITY_ID CHAR(15) NOT NULL > CONSTRAINT TEST_PK PRIMARY KEY ( > ORGANIZATION_ID, > SCORE DESC, > ENTITY_ID DESC > ) > ) VERSIONS=1, MULTI_TENANT=TRUE, REPLICATION_SCOPE=1, TTL=43200 > UPSERT INTO COMMUNITIES.TEST VALUES ('org1',3,'01'); > UPSERT INTO COMMUNITIES.TEST VALUES ('org1',2,'04'); > UPSERT INTO COMMUNITIES.TEST VALUES ('org1',2,'03'); > UPSERT INTO COMMUNITIES.TEST VALUES ('org1',1,'02'); > # QUERY # > SELECT entity_id, score > FROM communities.test > WHERE organization_id = 'org1' > AND (score, entity_id) < (2, '04') > ORDER BY score DESC, entity_id DESC > LIMIT 3 > I'm specifying the score and entity_id of the second row so I'm expecting it > to return the 2 last values (with id 03 and 02). However, it returns the 3 > first values (and if it wasn't for the LIMIT 3 it would just return all). -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (PHOENIX-3382) Query using Row Value Constructor comprised of non leading PK columns returns incorrect results
[ https://issues.apache.org/jira/browse/PHOENIX-3382?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15579001#comment-15579001 ] Samarth Jain commented on PHOENIX-3382: --- Looks good, +1. > Query using Row Value Constructor comprised of non leading PK columns returns > incorrect results > --- > > Key: PHOENIX-3382 > URL: https://issues.apache.org/jira/browse/PHOENIX-3382 > Project: Phoenix > Issue Type: Bug >Affects Versions: 4.8.0 >Reporter: Joel Palmert >Priority: Critical > Attachments: PHOENIX-3382.patch, PHOENIX-3382_wip.patch > > > I'm trying to use Row Value Constructors to do paging in a query in the below > table: > # TABLE # > CREATE TABLE IF NOT EXISTS COMMUNITIES.TEST ( > ORGANIZATION_ID CHAR(15) NOT NULL, > SCORE DOUBLE NOT NULL, > ENTITY_ID CHAR(15) NOT NULL > CONSTRAINT TEST_PK PRIMARY KEY ( > ORGANIZATION_ID, > SCORE DESC, > ENTITY_ID DESC > ) > ) VERSIONS=1, MULTI_TENANT=TRUE, REPLICATION_SCOPE=1, TTL=43200 > UPSERT INTO COMMUNITIES.TEST VALUES ('org1',3,'01'); > UPSERT INTO COMMUNITIES.TEST VALUES ('org1',2,'04'); > UPSERT INTO COMMUNITIES.TEST VALUES ('org1',2,'03'); > UPSERT INTO COMMUNITIES.TEST VALUES ('org1',1,'02'); > # QUERY # > SELECT entity_id, score > FROM communities.test > WHERE organization_id = 'org1' > AND (score, entity_id) < (2, '04') > ORDER BY score DESC, entity_id DESC > LIMIT 3 > I'm specifying the score and entity_id of the second row so I'm expecting it > to return the 2 last values (with id 03 and 02). However, it returns the 3 > first values (and if it wasn't for the LIMIT 3 it would just return all). -- This message was sent by Atlassian JIRA (v6.3.4#6332)