[
https://issues.apache.org/jira/browse/HBASE-14406?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14953360#comment-14953360
]
Ted Malaska commented on HBASE-14406:
-------------------------------------
I think the bug from last time was the following two
( rowkey < 1 or col > 2 )
and
( colA < 1 or colB > 2 )
The functionality of (rowkey < 1 and col > 2) worked in the last patch
But here are some related tests that should cover both cases
test("Test SQL point and range combo")
test("Test OR logic with a one RowKey and One column")
test("Test two complete range non merge rowKey query")
test("Test OR logic with a two columns")
> The dataframe datasource filter is wrong, and will result in data loss or
> unexpected behavior
> ---------------------------------------------------------------------------------------------
>
> Key: HBASE-14406
> URL: https://issues.apache.org/jira/browse/HBASE-14406
> Project: HBase
> Issue Type: Bug
> Components: spark
> Affects Versions: 2.0.0
> Reporter: Zhan Zhang
> Assignee: Ted Malaska
> Priority: Blocker
> Fix For: 2.0.0
>
> Attachments: HBASE-14406.1.patch, HBASE-14406.2.patch,
> HBASE-14406.3.patch
>
>
> Following condition will result in the same filter. It will have data loss
> with the current filter construction.
> col1 > 4 && col2 < 3
> col1 > 4 || col2 < 3
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
