[
https://issues.apache.org/jira/browse/CASSANDRA-2555?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13026322#comment-13026322
]
Jonathan Ellis commented on CASSANDRA-2555:
-------------------------------------------
{code}
if (up.isSetTimestamp())
throw new InvalidRequestException(
"Timestamp must be set on the BATCH, not
individual statements");
{code}
Unlike ConsistencyLevel, it can make sense to have different timestamps on the
individual updates. So we should only reject this if there is a ts on the batch
as well. (I.e., ts on batch = ok, ts on individual updates = ok, ts on both =
confusing and not ok.)
bq. 3 tests added
if it doesn't cause conflicts, can you move these into existing
batch/update/insert tests? There is a very high per-method overhead in
test_cql since it tears down, wipes, and restarts the server for each.
> Explicit timestamp support for CQL
> ----------------------------------
>
> Key: CASSANDRA-2555
> URL: https://issues.apache.org/jira/browse/CASSANDRA-2555
> Project: Cassandra
> Issue Type: Improvement
> Components: API
> Reporter: Jonathan Ellis
> Assignee: Pavel Yaskevich
> Fix For: 0.8.1
>
> Attachments: CASSANDRA-2555.patch
>
>
> To make CQL BATCHes useful for the normal purpose of idempotent retries, we
> need to add support for client-supplied timestamps. At a minimum this needs
> to be available for a batch (BEGIN BATCH USING TIMESTAMP 1303743619771318)
> but might also be nice at the statement level (UPDATE ... USING TIMESTAMP
> ...).
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
