[jira] [Commented] (CASSANDRA-7660) Indicate PK columns in "prepared" native protocol responses

["Benjamin Lerer (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/CASSANDRA-7660?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14349442#comment-14349442 ] 

Benjamin Lerer commented on CASSANDRA-7660:
-------------------------------------------

I noticed the following things:

* In {{VariableSpecifications.getPartitionKeyBindIndexes(CFMetaData cfm)}} the following line seems suspicious to me:
{{if (!targetColumn.ksName.equals(cfm.cfName) || !targetColumn.cfName.equals(cfm.ksName))}}

* The constructor {{Prepared(CQLStatement, VariableSpecifications)}} does not look to used anywhere.

* I feel that it will be more natural for {{allInSameCF(List<ColumnSpecification> names)}} to be in {{ColumnSpecification}} instead of in {{ResultMetadata}}. 

* I think that it would have been clearer to check for multiple cfm in {{BatchStatement}} rather than in {{VariableSpecification.getPartitionKeyBindIndexes}} even if it is probably less efficient. 

* I did not see any existing tests covering the serialization/deserialization of {{ResultMessage}} but I might have missed them. In case we do not have some it would be probably good to add some.

I had a look at the python driver code but I think that it would be good if someone else than me had a look at it as I am not that familiar with the driver.

Otherwise, I have found the change to the native protocol document quite clear.

> Indicate PK columns in "prepared" native protocol responses
> -----------------------------------------------------------
>
>                 Key: CASSANDRA-7660
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-7660
>             Project: Cassandra
>          Issue Type: Sub-task
>          Components: Core
>            Reporter: Tyler Hobbs
>            Assignee: Tyler Hobbs
>              Labels: protocolv4
>             Fix For: 3.0
>
>         Attachments: 7660-trunk.txt
>
>
> Currently, when native protocol drivers prepare statements, they need to check their internal representation of the schema to determine which, if any, of the columns are part of the partition key.  If the native protocol response indicated this, it would remove the drivers' dependency on having an internal representation of the schema when preparing statements.
> This would need to go in the v4 protocol.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)