[ https://issues.apache.org/jira/browse/CASSANDRA-3165?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Sylvain Lebresne updated CASSANDRA-3165: ---------------------------------------- Attachment: 3165_v2.patch We don't have to remember the factory, the columns can give it back to us. Adding v2 that does that. But actually v2 adds a cloneMeShallow variant that allows to specify the factory, which allow to change the backing map in the clone. The patch uses it in CFS.filterColumnFamily, which should provide some speedup in the cached case. > ArrayBackedSortedColumns is not being used by ColumnFamily.cloneMeShallow() > --------------------------------------------------------------------------- > > Key: CASSANDRA-3165 > URL: https://issues.apache.org/jira/browse/CASSANDRA-3165 > Project: Cassandra > Issue Type: New Feature > Components: Core > Affects Versions: 1.0 > Reporter: Yang Yang > Priority: Minor > Fix For: 1.1 > > Attachments: > 0001-ColumnFamily.cloneMeShallow-should-remember-the-fact.patch, 3165_v2.patch > > > JIRA 2843 utilizes ArrayBackedSortedColumns for read path. > but this is missed in ColumnFamily.cloneMeShallow(), which still calls > ColumnFamily.create() without the factory, so it defaults to the slower > SkipList-based one. > an easy change is to remember the factory, and use it when cloneMeShallow() -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira