[
https://issues.apache.org/jira/browse/HIVE-13380?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15321807#comment-15321807
]
Sergey Shelukhin commented on HIVE-13380:
-----------------------------------------
This results in unexpected (and arguably wrong) results in some queries (see
attached file).
In the query without decimal casts, 0.07 is not included in the range between
0.06-0.01 and 0.06+0.01 (it is included in the range between 0.05 and 0.07).
I think floating point types are an abomination that should never be used in
data systems unless explicitly called for... I wonder if we should revert this
before it's released in 2.1. cc [~jcamachorodriguez] just in case
At the very least we need to make sure that decimal is the default if the
column is decimal and non-column is double.
Thoughts?
> Decimal should have lower precedence than double in type hierachy
> -----------------------------------------------------------------
>
> Key: HIVE-13380
> URL: https://issues.apache.org/jira/browse/HIVE-13380
> Project: Hive
> Issue Type: Bug
> Components: Types
> Reporter: Ashutosh Chauhan
> Assignee: Ashutosh Chauhan
> Labels: TODOC2.1
> Fix For: 2.1.0
>
> Attachments: HIVE-13380.2.patch, HIVE-13380.4.patch,
> HIVE-13380.5.patch, HIVE-13380.patch, decimal_filter.q
>
>
> Currently its other way round. Also, decimal should be lower than float.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
