[
https://issues.apache.org/jira/browse/KUDU-2158?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16176982#comment-16176982
]
Dan Burkert commented on KUDU-2158:
-----------------------------------
> For compound keys isn't it important to also be able to specify an upper
> bound PK tuple?
Ah ok, it seems I misunderstood the original question. I thought this was about
pushing predicates onto individual columns. Indeed, there's no way to add an
inclusive upper primary key bound. It can be simulated by incrementing the
value of the final column in the PK, there's a non-public method in
[PartialRow|https://github.com/apache/kudu/blob/master/java/kudu-client/src/main/java/org/apache/kudu/client/PartialRow.java#L1093]
that does this, as an example.
Without some more evidence I don't think I'd be in favor of a built-in
inclusive upper bound method on the scan builder. Exclusive upper bound is
already very niche.
> AddUpperBound for KuduScanner
> -----------------------------
>
> Key: KUDU-2158
> URL: https://issues.apache.org/jira/browse/KUDU-2158
> Project: Kudu
> Issue Type: Improvement
> Reporter: Thomas Tauber-Marshall
> Priority: Critical
>
> There's work currently ongoing on the Impala side to support min-max runtime
> filters. To support that, it would be help if there was a
> KuduScanner::AddUpperBound() function.
> Currently, KuduScanTokenBuilder supports AddUpperBound() and KuduScanner
> support AddLowerBound() and AddExclusiveUpperBound(), but there's no
> inclusive KuduScanner::AddUpperBound()
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
