[
https://issues.apache.org/jira/browse/CASSANDRA-10820?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
DOAN DuyHai updated CASSANDRA-10820:
------------------------------------
Description:
Original JIRA created on driver side:
https://datastax-oss.atlassian.net/browse/JAVA-1008
After the refactoring of meta data, the index meta data has been moved to table
level to support different types of index (functional, composite ...), fine.
However, there is still a strong need to be able to associate an index with a
list of columns involved into its value. Given the example:
{code:sql}
CREATE CUSTOM INDEX my_index ON ks.t1(a, keys(b), foo(c)) USING 'indexclass'
{code}
We should be able to establish a 1:N relationship between the index and the
column:
* index -> a
* index -> b (with its key)
* index -> c (using UDF foo)
The original requirement is to able to display a table metadata and flag each
column to know whether it has an index or it or not (no matter it is a
simple/classic or composite index). I'm using this feature for *Apache Zeppelin
Cassandra interpreter* (DESCRIBE TABLE)
Ping [~beobal] because he worked on this with [~adutra]
was:
Original JIRA created on driver side:
https://datastax-oss.atlassian.net/browse/JAVA-1008
After the refactoring of meta data, the index meta data has been moved to table
level to support different types of index (functional, composite ...) fine.
However, there is still a strong need to be table to associate an index with a
list of columns involved into its value. Given the example:
{code:sql}
CREATE CUSTOM INDEX my_index ON ks.t1(a, keys(b), foo(c)) USING 'indexclass'
{code}
We should be able to establish a 1:N relationship between the index and the
column:
* index -> a
* index -> b (with its key)
* index -> c (using UDF foo)
The original requirement is to able to display a table metadata and flag each
column to know whether it has an index or it or not (no matter it is a
simple/classic or composite index). I'm using this feature for *Apache Zeppelin
Cassandra interpreter* (DESCRIBE TABLE)
> Enhance metadata for index to be able to link an index with 1 to N columns
> --------------------------------------------------------------------------
>
> Key: CASSANDRA-10820
> URL: https://issues.apache.org/jira/browse/CASSANDRA-10820
> Project: Cassandra
> Issue Type: Improvement
> Components: Distributed Metadata
> Environment: C* 3.0.x
> Reporter: DOAN DuyHai
>
> Original JIRA created on driver side:
> https://datastax-oss.atlassian.net/browse/JAVA-1008
> After the refactoring of meta data, the index meta data has been moved to
> table level to support different types of index (functional, composite ...),
> fine.
> However, there is still a strong need to be able to associate an index with a
> list of columns involved into its value. Given the example:
> {code:sql}
> CREATE CUSTOM INDEX my_index ON ks.t1(a, keys(b), foo(c)) USING 'indexclass'
> {code}
> We should be able to establish a 1:N relationship between the index and the
> column:
> * index -> a
> * index -> b (with its key)
> * index -> c (using UDF foo)
> The original requirement is to able to display a table metadata and flag each
> column to know whether it has an index or it or not (no matter it is a
> simple/classic or composite index). I'm using this feature for *Apache
> Zeppelin Cassandra interpreter* (DESCRIBE TABLE)
> Ping [~beobal] because he worked on this with [~adutra]
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
