[ 
https://issues.apache.org/jira/browse/CASSANDRA-11947?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16064226#comment-16064226
 ] 

ZhaoYang edited comment on CASSANDRA-11947 at 6/27/17 4:11 AM:
---------------------------------------------------------------

How about trying following schema? basically the same query as you want.
{code}
cqlsh:jira> CREATE MATERIALIZED VIEW analytics_ww AS
        ...        SELECT * FROM analytics
        ...        WHERE date IS NOT NULL AND resourceId IS NOT NULL AND 
country IS NOT NULL AND country = 'WW' 
        ...        PRIMARY KEY (resourceId, date, country);

{code}

The original idea to stop `collapsing` data in MV (part of base primary key 
columns in view) is to protect user from unexpected tombstones.. But in this 
case, if the `missing primary key in the view` is guarded by a filter 
condition, it should be safe.

If your use case can be done via above example, I suggest won'tfix. 


was (Author: jasonstack):
How about trying following schema? basically the same query as you want.
{code}
cqlsh:jira> CREATE MATERIALIZED VIEW analytics_ww AS
        ...        SELECT * FROM analytics
        ...        WHERE date IS NOT NULL AND resourceId IS NOT NULL AND 
country IS NOT NULL AND country = 'WW' 
        ...        PRIMARY KEY (resourceId, date, country);

{code}

The original idea to stop `collapsing` data in MV (part of base primary key 
columns in view) is to protect user from tombstones.. I think in this case, if 
the `missing primary key in the view` is guarded by a filter condition, it's 
safe.

If your use case can be done via above example, I suggest won'tfix. 

> Ability to create a Materialized View without primary key column that has '=' 
> relation
> --------------------------------------------------------------------------------------
>
>                 Key: CASSANDRA-11947
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-11947
>             Project: Cassandra
>          Issue Type: Improvement
>            Reporter: Nimi Wariboko Jr.
>
> I have a use case where I'd like to create a materialized view where one of 
> the columns is filtered on a '=' relation and the primary key for that 
> materialized view doesn't include that column.
> Example:
> {code}
> CREATE TABLE analytics (
>   resourceId text,
>   country text,
>   date timestamp,
>   clicks int,
>   PRIMARY KEY(resourceId, date, country)
> );
> CREATE MATERIALIZED VIEW analytics_ww AS
>        SELECT * FROM analytics
>        WHERE country = 'WW'
>        PRIMARY KEY (resourceId, date);
> {code}
> This way I can have a table filtered on a high cardinality column, and I 
> don't also have to specify that column, even though I already know thats the 
> only value in the view.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org

Reply via email to