[ https://issues.apache.org/jira/browse/CASSANDRA-1480?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12908851#action_12908851 ]
Jeremy Hanna commented on CASSANDRA-1480: ----------------------------------------- I kind of wondered if it would be complicated like that based on how thrift handled it. It's a one line change to fix it in ColumnFamilyRecordReader. I didn't know if it would affect anything else similarly. Hopefully system/unit tests catch a lot of that. We could do some simple searches in the code for null checks I suppose - and in those cases, if they're strings, do a o.a.commons.lang.StringUtils.isNotBlank instead. Not sure if it's worth it though since that could have other side effects. > CFMetaData.convertToThrift makes subcomparator_type empty string instead of > null > -------------------------------------------------------------------------------- > > Key: CASSANDRA-1480 > URL: https://issues.apache.org/jira/browse/CASSANDRA-1480 > Project: Cassandra > Issue Type: Bug > Components: API > Affects Versions: 0.7 beta 1 > Reporter: Jeremy Hanna > Assignee: Jon Hermes > Fix For: 0.7 beta 2 > > > As a result of CASSANDRA-891 adding a CFMetaData.convertToThrift method, the > values such as subcomparator_type are defaulted to empty string instead of > null. That makes it so, for example, in ColumnFamilyRecordReader, in its > RowIterator, the check for only null is insufficient. It also needs to check > for a blank value. > After a discussion about it in IRC, Jonathan said it was probably easier to > just change the creation to give a null value instead of empty string. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.