[
https://issues.apache.org/jira/browse/CASSANDRA-1566?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12916902#action_12916902
]
Jesse McConnell commented on CASSANDRA-1566:
--------------------------------------------
is it not wrong to have a user expecting to be creating a secondary index, have
put everything in place to make that happen and have cassandra silently _not_
do the intended behavior, to have an optional parameter deceptively named
index_type that really governs if a secondary index gets made or not?
based on the logic being implemented here info.thrift_type would be better
renamed to be a boolean 'info.make_index' and have it be optionally defined as
true if you actually want an index
> indexType required for secondary indices and fails silently with out it
> -----------------------------------------------------------------------
>
> Key: CASSANDRA-1566
> URL: https://issues.apache.org/jira/browse/CASSANDRA-1566
> Project: Cassandra
> Issue Type: Bug
> Components: API
> Affects Versions: 0.7 beta 2
> Reporter: Jesse McConnell
> Priority: Minor
>
> When defining an secondary index on a column family the indexType is required
> to be set to KEYS and if you don't set that cassandra will silently fail to
> add the index
> for (ColumnDefinition info : metadata.column_metadata.values())
> {
> if (info.index_type != null)
> addIndex(info);
> }
> You are able to make it to this code in the ColumnFamilyStore with no
> warnings (that I noticed at least) and there ought to at least be an info
> message here where it indicates the index is being tossed during creation.
> Better yet there ought to be protection on the thrift call that requires this
> option for the request to be valid or default it to the only option right now
> of KEYS
> cheers
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
