[jira] [Updated] (DRILL-2739) Drill stops reading columns > 4095

["Nick Amato (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/DRILL-2739?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Nick Amato updated DRILL-2739:
------------------------------
    Attachment: mcols.csv

This file contains 1 row of 10000 comma-separated integers (columns).  I encountered the issue with a file with 20000 rows of the same data, but was able to reproduce it with this 1-line file to save some space...

> Drill stops reading columns > 4095
> ----------------------------------
>
>                 Key: DRILL-2739
>                 URL: https://issues.apache.org/jira/browse/DRILL-2739
>             Project: Apache Drill
>          Issue Type: Bug
>          Components: Execution - Flow, Storage - Text & CSV
>    Affects Versions: 0.7.0
>         Environment: CentOS 6.4
> sqlline 1.1.6
> Drill 0.7
>            Reporter: Nick Amato
>            Assignee: Chris Westin
>         Attachments: mcols.csv
>
>
> Querying a CSV file with 10000 columns, it looks like I can't query past column 4095. 
> 0: jdbc:drill:localhost:5181> SELECT columns[4094] FROM `dfs`.`tmp`.`manycols.csv` limit 1;
> +------------+
> |   EXPR$0   |
> +------------+
> | 833        |
> +------------+
> 1 row selected (0.269 seconds)
> 0: jdbc:drill:localhost:5181> SELECT columns[4095] FROM `dfs`.`tmp`.`manycols.csv` limit 1;
> +--+
> |  |
> +--+
> +--+
> No rows selected (0.198 seconds)
> 0: jdbc:drill:localhost:5181>
> [mapr@drilltwitter tmp]$ awk -F, '{ print $4095,$4096,$4097 }' manycols.csv | head -1
> 833 2552 2222
> [mapr@drilltwitter tmp]$



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)