drill-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Kunal Khatua (JIRA)" <j...@apache.org>
Subject [jira] [Closed] (DRILL-4812) Wildcard queries fail on Windows
Date Wed, 29 Mar 2017 22:56:41 GMT

     [ https://issues.apache.org/jira/browse/DRILL-4812?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Kunal Khatua closed DRILL-4812.
-------------------------------

Verified on Windows 10 by creating a nested table of nation  (25row parquet file) with a directory
structure of 2 level-1 dirs (a,b), 3 level-2 dirs (1,2,3) and 3 level-3 dirs (yes,no, idk).

{code}
0: jdbc:drill:zk=local> select count(n_nationkey) from dfs.root.`/drill/nation/a/1/idk/part-m-00000.parquet`;
+---------+
| EXPR$0  |
+---------+
| 25      |
+---------+
1 row selected (0.178 seconds)
0: jdbc:drill:zk=local> select count(n_nationkey) from dfs.root.`/drill/nation/a/1/idk`;
+---------+
| EXPR$0  |
+---------+
| 25      |
+---------+
1 row selected (0.169 seconds)
0: jdbc:drill:zk=local> select count(n_nationkey) from dfs.root.`/drill/nation/a/*/idk`;
+---------+
| EXPR$0  |
+---------+
| 75      |
+---------+
1 row selected (0.167 seconds)
0: jdbc:drill:zk=local> select count(n_nationkey) from dfs.root.`/drill/nation/*/*/idk`;
+---------+
| EXPR$0  |
+---------+
| 150     |
+---------+
1 row selected (0.226 seconds)
0: jdbc:drill:zk=local> select count(n_nationkey) from dfs.root.`/drill/nation/*/*/*`;
+---------+
| EXPR$0  |
+---------+
| 450     |
+---------+
1 row selected (0.225 seconds)
{code}

> Wildcard queries fail on Windows
> --------------------------------
>
>                 Key: DRILL-4812
>                 URL: https://issues.apache.org/jira/browse/DRILL-4812
>             Project: Apache Drill
>          Issue Type: Bug
>          Components: Storage - Other
>    Affects Versions: 1.7.0
>         Environment: Windows 7
>            Reporter: Mike Lavender
>              Labels: easyfix, easytest, ready-to-commit, windows
>             Fix For: 1.10.0
>
>
> Wildcards within the path of a query are not handled on windows and result in a "String
index out of range" exception.
> for example:
> {noformat}
> 0: jdbc:drill:zk=local> SELECT SUM(qty) as num FROM dfs.parquet.`/trends/2016/1/*/*/3701`;
> Error: VALIDATION ERROR: String index out of range: -1
> SQL Query null
> {noformat}
> ----
> The problem exists within:
> exec\java-exec\src\main\java\org\apache\drill\exec\store\dfs\FileSelection.java
> private static Path handleWildCard(final String root)
> This function is looking for the index of the system specific PATH_SEPARATOR which on
windows is '\' (from System.getProperty("file.separator")).  The path passed in to handleWildcard
will not ever have those type of path separators as the Path constructor (from org.apache.hadoop.fs.Path)
sets all the path separators to '/'.
> NOTE:
> private static String removeLeadingSlash(String path)
> in that same file explicitly looks for '/' and does not use the system specific PATH_SEPARATOR.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

Mime
View raw message