[ 
https://issues.apache.org/jira/browse/CASSANDRA-6839?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14037389#comment-14037389
 ] 

Sylvain Lebresne commented on CASSANDRA-6839:
---------------------------------------------

Mostly looks good, a few minor remarks:
* Should include operator in ColumnCondition.Bound hashCode methods.
* In ColumnCondition.setAppliesTo and mapAppliesTo, the Set/Map in parameter 
are already properly sorted (Sets/Maps.Value make sure of that) so we can 
iterate on them directly and save the allocation/copy (the previous logic was 
creating new TreeSet/TreeMap for other reasons that were misguided).
* Note sure why CFDefinition.Name is made @VisibleForTesting?
* What's the reason for removing the lines in 
MultiColumnRelationTest.testMultipleClusteringColumnInequalityReversedComponents.

bq. It seems like we should also add support for IN. I can add do that in this 
ticket, if we'd like, or I can create a new ticket for that.

Agreed that it's worth adding. I don't mind if we move that to a follow up 
ticket, though if we do then I think we need to validate it's currently not 
used for conditions (I don't think it's done currently and so an AssertionError 
will be thrown). Maybe it's easier to just go ahead and add support to this 
patch directly (unless it adds too much complexity but it sounds like it 
shouldn't)?


> Support non equal conditions (for LWT)
> --------------------------------------
>
>                 Key: CASSANDRA-6839
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-6839
>             Project: Cassandra
>          Issue Type: Improvement
>            Reporter: Sylvain Lebresne
>            Assignee: Tyler Hobbs
>            Priority: Minor
>             Fix For: 2.0.9
>
>         Attachments: 6839.txt
>
>
> We currently only support equal conditions in conditional updates, but it 
> would be relatively trivial to support non-equal ones as well. At the very 
> least we should support '>', '>=', '<' and '<=', though it would probably 
> also make sense to add a non-equal relation too ('!=').



--
This message was sent by Atlassian JIRA
(v6.2#6252)

Reply via email to