You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@drill.apache.org by "Zhiyong Liu (JIRA)" <ji...@apache.org> on 2014/05/19 19:38:38 UTC
[jira] [Commented] (DRILL-495) Drill does not return columns in
appropriate order for select all and aggregate selects
[ https://issues.apache.org/jira/browse/DRILL-495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14002039#comment-14002039 ]
Zhiyong Liu commented on DRILL-495:
-----------------------------------
Updates:
This has been fixed for the following query:
select * from voter where age <=20;
But column order issue also happens with the following query:
select p.p_mfgr as mfgr, count(*) as mycount
from part p
where p.p_size < 100
and p.p_type like '%NICKEL'
group by p.p_mfgr
order by 1
on build:
git.commit.id.abbrev=70fab8c
git.commit.id=70fab8c968a6dc05471aa1f32009cd15945d4f3d
> Drill does not return columns in appropriate order for select all and aggregate selects
> ---------------------------------------------------------------------------------------
>
> Key: DRILL-495
> URL: https://issues.apache.org/jira/browse/DRILL-495
> Project: Apache Drill
> Issue Type: Bug
> Affects Versions: 1.0.0
> Environment: CentOS release 6.5
> Reporter: Krystal
>
> I have a parquet file with the following columns:
> rownum name age registration contributions voterzone create_time
> I ran the following query via jdbc/sqlline:
> select * from voter where age <=20;
> I would expect the columns returned are the same as the order in the parquet data file. However the returned result is as follows:
> 0: jdbc:drill:schema=dfs> select * from voter where age <=20;
> +------------+------------+---------------+------------+------------+--------------+-------------+
> | rownum | age | contributions | voterzone | name | registration | create_time |
> +------------+------------+---------------+------------+------------+--------------+-------------+
> | 3 | 18 | 128.2 | 8750 | [B@3a942c3a | [B@55e3b1e1 | [B@57b356d4 |
> | 22 | 19 | 16.25 | 27833 | [B@3f09a547 | [B@1241f8a6 | [B@292b63a7 |
> | 57 | 19 | 265.9 | 11041 | [B@5f9b7e0e | [B@5970fa2b | [B@8384aed |
> | 59 | 18 | 835.31 | 11276 | [B@27bf11d2 | [B@712b0660 | [B@3fafc2ab |
> | 60 | 20 | 53.19 | 7506 | [B@2c158937 | [B@39e3907d | [B@b231c3c |
> | 70 | 18 | 94.03 | 12853 | [B@2e12acda | [B@4c1233d7 | [B@3f098f45 |
> | 85 | 19 | 497.94 | 8981 | [B@c9d1b58 | [B@1e6e34e0 | [B@55516dbb |
> I observed the same column ordering problem when using aggregates in the select statement. For example:
> 0: jdbc:drill:schema=dfs> select name as col1, sum(contributions) as col2 from voter group by name order by name;
> +------------+------------+
> | col2 | col1 |
> +------------+------------+
> | 412.64 | [B@e2d9423 |
> | 777.23 | [B@52588d1a |
> | 4024.16 | [B@3397d032 |
> | 1417.0300000000002 | [B@33e59d14 |
> The columns returned should be col1 col2.
--
This message was sent by Atlassian JIRA
(v6.2#6252)