[
https://issues.apache.org/jira/browse/IGNITE-26613?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Andrey Mashenkov updated IGNITE-26613:
--------------------------------------
Description:
For now, in come cases queries, which uses `{{{}FLOAT`{}}} or `DOUBLE`, may
return inconsistent results for {{`NaN`}} values.
* `{{{}SELECT MAX(f)`{}}} returns {{`NaN`}} even if there is another non-null
values.
* `{{{}SELECT MIN(f)`{}}} returns minimal value if there smth else except
{{`NaN`.}}
* `{{{}SELECT f ... WHERE f > 0`{}}} omit {{NaN}} values as well as `nulls`.
* `SELECT f .. GROUP BY f` returns a single group for `NaN` values.
Let's validate this behavior and fix if needed, or explain why all results are
valid.
was:
For now, in come cases queries, which uses `{{{}FLOAT`{}}} or `DOUBLE`, may
return inconsistent results for {{`NaN`}} values.
* `{{{}SELECT MAX(f)`{}}} returns {{`NaN`}}
* `{{{}SELECT f ... WHERE f > 0`{}}} omit {{NaN}} values.
* `SELECT f .. GROUP BY f` returns a single group for `NaN` values.
Let's validate this behavior and fix if needed, or explain why all results are
valid.
> Sql. Fix SQL queries results inconsistency for NaN values.
> ----------------------------------------------------------
>
> Key: IGNITE-26613
> URL: https://issues.apache.org/jira/browse/IGNITE-26613
> Project: Ignite
> Issue Type: Bug
> Components: sql ai3
> Reporter: Andrey Mashenkov
> Priority: Major
> Labels: ignite-3
>
> For now, in come cases queries, which uses `{{{}FLOAT`{}}} or `DOUBLE`, may
> return inconsistent results for {{`NaN`}} values.
> * `{{{}SELECT MAX(f)`{}}} returns {{`NaN`}} even if there is another
> non-null values.
> * `{{{}SELECT MIN(f)`{}}} returns minimal value if there smth else except
> {{`NaN`.}}
> * `{{{}SELECT f ... WHERE f > 0`{}}} omit {{NaN}} values as well as `nulls`.
> * `SELECT f .. GROUP BY f` returns a single group for `NaN` values.
> Let's validate this behavior and fix if needed, or explain why all results
> are valid.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
