[ https://issues.apache.org/jira/browse/CASSANDRA-14099?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Vincent White updated CASSANDRA-14099: -------------------------------------- Status: Patch Available (was: Open) I've created a patch that splits in this comparator in two as a way to maybe help avoid this confusion in the future. Now that this is split into two I'm not sure if a unit test for ageSortedSSTables (or the comparators themselves) would be required? I have included a unit test for ageSortedSSTables on my 3.0.x branch, not sure if it's worth making ageSortedSStable() Public just for this but I didn't see anywhere else where its behaviour was visible. [3.0 patch | https://github.com/vincewhite/cassandra/commits/14099_timestamp_comparators_30] [3.0 utest | https://github.com/vincewhite/cassandra/commit/5ab1ff36a28b41039bd93de7d47b4131e1c2dfaa] [3.x patch | https://github.com/vincewhite/cassandra/commits/14099_timestamp_comparators_311] [trunk patch | https://github.com/vincewhite/cassandra/commits/14099_timestamp_comparators_trunk] > LCS ordering of sstables by timestamp is inverted > ------------------------------------------------- > > Key: CASSANDRA-14099 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14099 > Project: Cassandra > Issue Type: Bug > Components: Compaction > Reporter: Jeff Jirsa > Priority: Minor > Fix For: 3.0.x, 3.11.x, 4.x > > > In CASSANDRA-14010 we discovered that CASSANDRA-13776 broke sstable ordering > by timestamp (inverted it accidentally). Investigating that revealed that the > comparator was expecting newest-to-oldest for read command, but LCS expects > oldest-to-newest. -- This message was sent by Atlassian JIRA (v6.4.14#64029) --------------------------------------------------------------------- To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org