[
https://issues.apache.org/jira/browse/ACCUMULO-956?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13551295#comment-13551295
]
Brian Loss commented on ACCUMULO-956:
-------------------------------------
I not sure there is a problem. The KeyTransformingIterator transforms only
single keys, so you can't do any aggregation. If you had some kind of
aggregating iterator on top of it on the stack, it would need to read enough
keys from its source to do the aggregation. To produce your example, I think
you'd have to scan all the cf: keys from the source to know it has what it
needs for r cf: even [v2, v4]. Since that iterator would be scanning its
source, there is no problem with the stack getting torn down and re-seeked. Or
am I missing something?
> Iterator to transform key parts
> -------------------------------
>
> Key: ACCUMULO-956
> URL: https://issues.apache.org/jira/browse/ACCUMULO-956
> Project: Accumulo
> Issue Type: Improvement
> Reporter: Brian Loss
> Fix For: 1.5.0
>
> Attachments: key_transforming_iterator.patch
>
>
> Iterators that transform parts of the key can be tricky if any transformation
> affects sort ordering. Implement an iterator that takes care of the tricky
> details that come with modifying sort order (e.g., handling scan-time
> iterator reconstruction and the associated seek).
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
