flink-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Dawid Wysakowicz (Jira)" <j...@apache.org>
Subject [jira] [Commented] (FLINK-15602) Blink planner does not respect the precision when casting timestamp to varchar
Date Thu, 16 Jan 2020 09:06:00 GMT

    [ https://issues.apache.org/jira/browse/FLINK-15602?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17016723#comment-17016723
] 

Dawid Wysakowicz commented on FLINK-15602:
------------------------------------------

[~docete] Thank you for checking it.

I checked additionally INTERVAL and DECIMAL types and SQL Server pads both types, Oracle does
not pad DECIMAL.
Personally I am in favor of padding. [~twalthr] gave also a nice argument that padded values
look nicer when there are multiple values being printed:

{code}
// Padded version
2014-07-02 06:14:00.010000000
2014-07-02 06:14:00.001000000
2014-07-02 06:14:00.001000001

// Not padded

2014-07-02 06:14:00.01
2014-07-02 06:14:00.001
2014-07-02 06:14:00.001000001
{code}

MySQL and Hive/Spark are not known to follow the SQL too closely so I would not take them
as good examples.

If we decide not to pad the results we should fix the {{sql client kafka tests}}

> Blink planner does not respect the precision when casting timestamp to varchar
> ------------------------------------------------------------------------------
>
>                 Key: FLINK-15602
>                 URL: https://issues.apache.org/jira/browse/FLINK-15602
>             Project: Flink
>          Issue Type: Bug
>          Components: Table SQL / Planner
>    Affects Versions: 1.10.0
>            Reporter: Dawid Wysakowicz
>            Priority: Blocker
>             Fix For: 1.10.0
>
>
> According to SQL 2011 Part 2 Section 6.13 General Rules 11) d)
> {quote}
> If SD is a datetime data type or an interval data type then let Y be the shortest character
string that
> conforms to the definition of <literal> in Subclause 5.3, “<literal>”,
and such that the interpreted value
> of Y is SV and the interpreted precision of Y is the precision of SD.
> {quote}
> That means:
> {code}
> select cast(cast(TO_TIMESTAMP('2014-07-02 06:14:00', 'YYYY-MM-DD HH24:mm:SS') as TIMESTAMP(0))
as VARCHAR(256)) from ...;
> // should produce
> // 2014-07-02 06:14:00
> select cast(cast(TO_TIMESTAMP('2014-07-02 06:14:00', 'YYYY-MM-DD HH24:mm:SS') as TIMESTAMP(3))
as VARCHAR(256)) from ...;
> // should produce
> // 2014-07-02 06:14:00.000
> select cast(cast(TO_TIMESTAMP('2014-07-02 06:14:00', 'YYYY-MM-DD HH24:mm:SS') as TIMESTAMP(9))
as VARCHAR(256)) from ...;
> // should produce
> // 2014-07-02 06:14:00.000000000
> {code}
> One possible solution would be to propagate the precision in {{org.apache.flink.table.planner.codegen.calls.ScalarOperatorGens#localTimeToStringCode}}.
If I am not mistaken this problem was introduced in [FLINK-14599]



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Mime
View raw message