hive-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Fei Hui (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HIVE-15849) hplsql should add enterGlobalScope func to UDF
Date Tue, 14 Mar 2017 15:30:41 GMT

    [ https://issues.apache.org/jira/browse/HIVE-15849?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15924415#comment-15924415
] 

Fei Hui commented on HIVE-15849:
--------------------------------

[~ashutoshc] Could you please manually trigger the Jenkins job for this issue? Maybe PreCommitAdmin
does not work, but  i have no login privileges to builds@apache. Thanks

> hplsql should add enterGlobalScope func to UDF
> ----------------------------------------------
>
>                 Key: HIVE-15849
>                 URL: https://issues.apache.org/jira/browse/HIVE-15849
>             Project: Hive
>          Issue Type: Bug
>          Components: hpl/sql
>    Affects Versions: 2.2.0
>            Reporter: Fei Hui
>            Assignee: Fei Hui
>         Attachments: HIVE-15849.1.patch, HIVE-15849.patch
>
>
> code in Udf.java
> {code:title=Udf.java|borderStyle=solid}
>     if (exec == null) {
>       exec = new Exec();
>       String query = queryOI.getPrimitiveJavaObject(arguments[0].get());
>       String[] args = { "-e", query, "-trace" };
>       try {
>         exec.setUdfRun(true);
>         exec.init(args);
>       } catch (Exception e) {
>         throw new HiveException(e.getMessage());
>       }
>     }
>     if (arguments.length > 1) {
>       setParameters(arguments);
>     }
>     Var result = exec.run();
>     if (result != null) {
>       return result.toString();
>     }
> {code}
> Here is my thoughts
> {quote}
> we should add 'exec.enterGlobalScope(); '  between 'exec = new Exec();' and 'setParameters(arguments);'
> Because if we do not call exec.enterGlobalScope(),  setParameters(arguments) will useless.
Vars are not added into scope , but exec.run() will use vars which we set. The vars are parameters
passed to UDF, [, :1, :2, ...n] which are description in Udf.java
> {quote}
> Before add this function, the result as follow. we get the wrong result, because the
result contains  empty string  
> {quote}
> Starting pre-SQL statement
> Starting pre-SQL statement
> Starting pre-SQL statement
> Starting pre-SQL statement
> Starting pre-SQL statement
> Starting query
> Query executed successfully (2.30 sec)
> Ln:8 SELECT completed successfully
> Ln:8 Standalone SELECT executed: 1 columns in the result set
> Hello, !
> Hello, !
> {quote}
> After add this function, we get the right result
> {quote}
> Starting pre-SQL statement
> Starting pre-SQL statement
> Starting pre-SQL statement
> Starting pre-SQL statement
> Starting pre-SQL statement
> Starting query
> Query executed successfully (2.35 sec)
> Ln:8 SELECT completed successfully
> Ln:8 Standalone SELECT executed: 1 columns in the result set
> Hello, fei!
> Hello, fei!
> {quote}
> tests come from http://www.hplsql.org/udf



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

Mime
View raw message