[ https://issues.apache.org/jira/browse/CASSANDRA-6119?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13783806#comment-13783806 ]
Fabien Rousseau commented on CASSANDRA-6119: -------------------------------------------- After having reviewed your patch, it has definitely a better approach minimising code change and better readability So I'm +1 for your patch > IndexedSliceReader can skip columns when fetching multiple contiguous slices > ---------------------------------------------------------------------------- > > Key: CASSANDRA-6119 > URL: https://issues.apache.org/jira/browse/CASSANDRA-6119 > Project: Cassandra > Issue Type: Bug > Reporter: Fabien Rousseau > Assignee: Fabien Rousseau > Attachments: 6119.patch, 6119-v2.txt > > > This was observed using SliceQueryFilter with multiple slices. > Let's have a row "a" having the following column list : "colA", "colB", > "colC", "colD" > Then select 2 ranges : ["colA", "colB"], ["colC", "colD"] > Expected result is the four columns > But only 3 are returned ("colA", "colB", "colD") > To reproduce the above scenario in the unit tests, you can modify the test > "ColumnFamilyStoreTest.testMultiRangeIndexed" by replacing the original line : > String[] letters = new String[] { "a", "b", "c", "d", "e", "f", "g", > "h", "i" }; > by this one ("f" letter has been removed) : > String[] letters = new String[] { "a", "b", "c", "d", "e", "g", "h", > "i" }; > Anyway, a patch is attached which adds more unit tests, and modifies > IndexedSliceReader.IndexedBlockFetcher & > IndexedSliceReader.SimpleBlockFetcher -- This message was sent by Atlassian JIRA (v6.1#6144)