[
https://issues.apache.org/jira/browse/DRILL-2906?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14529331#comment-14529331
]
Steven Phillips commented on DRILL-2906:
----------------------------------------
This seems to be a problem with reading extended json, and is not related to
the writer.
For example, here are two files, where a.json uses extended json, but b.json
does not:
{code}
$ cat /tmp/a.json
{
"int_col" : {"$numberLong": 1},
"date_col" : {"$dateDay": "2012-05-22"},
"time_col" : {"$time": "19:20:30.45Z"}
}
$ cat /tmp/b.json
{
"int_col" : 1,
"date_col" : "2012-05-22",
"time_col" : "19:20:30.45Z"
}
{code}
Here is the output when querying:
{code}
0: jdbc:drill:> select * from dfs.tmp.`a.json`;
+------------+------------+------------+------------+
| int_col | date_col | time_col | * |
+------------+------------+------------+------------+
| 1 | 2012-05-22 | 11:20:30.450 | null |
+------------+------------+------------+------------+
1 row selected (1.477 seconds)
0: jdbc:drill:> select * from dfs.tmp.`b.json`;
+------------+------------+------------+
| int_col | date_col | time_col |
+------------+------------+------------+
| 1 | 2012-05-22 | 19:20:30.45Z |
+------------+------------+------------+
1 row selected (0.222 seconds)
{code}
> CTAS with store.format = 'json' returns incorrect results
> ---------------------------------------------------------
>
> Key: DRILL-2906
> URL: https://issues.apache.org/jira/browse/DRILL-2906
> Project: Apache Drill
> Issue Type: Bug
> Components: Storage - JSON, Storage - Writer
> Reporter: Mehant Baid
> Assignee: Steven Phillips
> Fix For: 1.0.0
>
>
> Performing a CTAS with 'store.format' = 'json' and querying the table results
> in projecting an addition field '*' will null values. Below is a simple repro
> 0: jdbc:drill:zk=local> create table t as select timestamp '1980-10-01
> 00:00:00' from cp.`employee.json` limit 1;
> +------------+---------------------------+
> | Fragment | Number of records written |
> +------------+---------------------------+
> | 0_0 | 1 |
> +------------+---------------------------+
> 1 row selected (0.314 seconds)
> 0: jdbc:drill:zk=local> select * from t;
> +------------+------------+
> | EXPR$0 | * |
> +------------+------------+
> | 1980-10-01 00:00:00.0 | null |
> +------------+------------+
> Notice in the above result set we get an extra column '*' with null value.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
