[
https://issues.apache.org/jira/browse/CASSANDRA-13186?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15970410#comment-15970410
]
Aleksandr Sorokoumov commented on CASSANDRA-13186:
--------------------------------------------------
In order to better understand the issue, I have been reading how secondary
indexes work and would like to clarify one point before submitting the patch.
>From the documentation
>(http://docs.datastax.com/en/cql/3.3/cql/cql_reference/cqlCreateIndex.html):
{quote}
Cassandra supports creating an index on most columns, excluding counter columns
but including a clustering column of a compound primary key or on the partition
(primary) key itself
{quote}
It is not clear to me how a secondary index on the whole partition key (not
just a component) works. AFAIU the index is implemented by maintaining a hidden
table where the indexed column is the primary key and the partition key of the
original table is the value. (Source:
https://www.datastax.com/dev/blog/cassandra-native-secondary-index-deep-dive).
In the hypothetical case, if the column is both the primary key and the native
secondary index, both columns of the hidden table would contain the same values.
> Create index fails if the primary key is included, but docs claim it is
> supported
> ---------------------------------------------------------------------------------
>
> Key: CASSANDRA-13186
> URL: https://issues.apache.org/jira/browse/CASSANDRA-13186
> Project: Cassandra
> Issue Type: Bug
> Components: CQL
> Reporter: Ariel Weisberg
> Assignee: Aleksandr Sorokoumov
>
> {noformat}
> cqlsh:foo> CREATE TABLE users (
> ... userid text PRIMARY KEY,
> ... first_name text,
> ... last_name text,
> ... emails set<text>,
> ... top_scores list<int>,
> ... todo map<timestamp, text>
> ... );
> cqlsh:foo> create index bar on foo.users (userid, last_name);
> InvalidRequest: Error from server: code=2200 [Invalid query] message="Cannot
> create secondary index on partition key column userid"
> {noformat}
> {quote}
> <beobal> yes, it's a bug in CreateIndexStatement. The check to enforce the PK
> has only a single component is wrong
> it considers each target in isolation, so it doesn't take into account that
> you might be creating a custom index on a PK component + something else
> {quote}
> http://docs.datastax.com/en/cql/3.3/cql/cql_reference/cqlCreateIndex.html
> {quote}
> Cassandra supports creating an index on most columns, excluding counter
> columns but including a clustering column of a compound primary key or on the
> partition (primary) key itself.
> {quote}
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
