[
https://issues.apache.org/jira/browse/HIVE-2738?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13196216#comment-13196216
]
Nicolas Lalevée commented on HIVE-2738:
---------------------------------------
Actually there was some classpath issues, the version of jackson use by hadoop
and mine are conflicting. The result is that my genericUDF failed to be
instanciated, and the XMLDecoder, rather than reporting the error, returns
null. Hence the NPE.
I suggest the XMLDecoder should always fail on error. See the attached patch.
> NPE in ExprNodeGenericFuncEvaluator
> -----------------------------------
>
> Key: HIVE-2738
> URL: https://issues.apache.org/jira/browse/HIVE-2738
> Project: Hive
> Issue Type: Bug
> Affects Versions: 0.8.0
> Reporter: Nicolas Lalevée
> Attachments: 750c8966-6402-465a-b011-903469fe56da.xml,
> HIVE-2738-r1237763.patch, MapMaxUDF.java, MapToJsonUDF.java, hive_job_logs.txt
>
>
> Here is the query:
> bq. {{SELECT t.lid, '2011-12-12',
> s_map2json(s_maxmap(UNION_MAP(t.categoryCount), 100)) FROM ( SELECT theme_lid
> AS theme_lid, MAP(s_host(referer), COUNT( * )) AS categoryCount FROM
> PageViewEvent WHERE day >= '20130104' AND day <= '20130112' AND date_ >=
> '2012-01-04' AND date_ < '2012-01-13' AND lid IS NOT NULL GROUP BY lid,
> s_host(referer) ) t GROUP BY t.lid}}
> Removing the call s_map2json make it work but not by removing s_maxmap, but I
> don't understand what could be wrong with the implementation of my udf. And I
> don't know how to debug remote hadoop jobs.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
