[ https://issues.apache.org/jira/browse/CASSANDRA-953?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Stu Hood updated CASSANDRA-953: ------------------------------- Attachment: 0003-Add-ByteOrderedPartitioner.patch 0002-Convert-Thrift-to-byte-keys.patch 0001-byte-for-decorateKey-and-getToken-in-IPartitioner.patch 0001 is super chunky (apologies), but in summary, it converts things like RowMutation, Row, Filters, and their accompanying tests to byte[] keys. 0002 is just the Thrift changes. (note, no Avro changes here: see CASSANDRA-958) 0003 adds a ByteOrderedPartitioner. > Expose byte[] row keys in client APIs > ------------------------------------- > > Key: CASSANDRA-953 > URL: https://issues.apache.org/jira/browse/CASSANDRA-953 > Project: Cassandra > Issue Type: Sub-task > Reporter: Stu Hood > Assignee: Stu Hood > Fix For: 0.7 > > Attachments: > 0001-byte-for-decorateKey-and-getToken-in-IPartitioner.patch, > 0002-Convert-Thrift-to-byte-keys.patch, 0003-Add-ByteOrderedPartitioner.patch > > > Once 767 lands, row keys will be stored as bytes internally, and the > partitioner will decide whether to interpret the bytes. We can then expose > the byte[] keys to client APIs. > One key issue to settle is whether we should support both keys and bytes for > the 0.7 release, or if we should make 0.7 a fully breaking release. The main > problem with supporting both key types is that I don't think Thrift supports > method overloading, so all of the methods would need to change names. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.