[ https://issues.apache.org/jira/browse/SOLR-12251?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Shawn Heisey reopened SOLR-12251: --------------------------------- On a closer look, I see exactly what you're talking about in DocBuilder. Using HashSet, the order of the results is lost. I've cooked up a patch that switches all the usages of HashSet to LinkedHashSet, and also eliminates all warnings noticed by my IDE (eclipse). > pk ids not sort when in deltaQuery > ----------------------------------- > > Key: SOLR-12251 > URL: https://issues.apache.org/jira/browse/SOLR-12251 > Project: Solr > Issue Type: Bug > Security Level: Public(Default Security Level. Issues are Public) > Components: contrib - DataImportHandler > Affects Versions: 7.1 > Environment: windows10 > Solr7.1 > JDK8 > > Reporter: wzhonggo > Priority: Major > > I use solr and mysql for search. > > {code:xml} > // data-config.xml > query="select * from score order by create_date asc" > deltaImportQuery="select * from score where id='${dih.delta.id}'" > deltaQuery="selectid from score where update_date > > '${dataimporter.last_index_time}' order by create_date asc " > {code} > > Mysql has three rows data in *score* table > > ||id||name||score||create_date||update_date|| > |UUID1|user1|60|2018-04-10|2018-04-10| > |UUID2|user1|70|2018-04-11 |2018-04-11| > |UUID3|user1|80|2018-04-12|2018-04-12| > The expected results In solr doc > ||Name||Score||CreateDate||UpdateDate|| > |user1|80|2018-04-12|2018-04-12| > > Use full import it will correct , but use delta import will wrong. > In the *org.apache.solr.handler.dataimport.DocBuilder* class , return not > LinkHashSet in > *collectDelta* method. > > Thanks. > > > > -- This message was sent by Atlassian JIRA (v7.6.3#76005) --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org