[jira] [Commented] (DRILL-2765) Throwing a NPE on exceuting query against parquet files

["Masood Parvari (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/DRILL-2765?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14493633#comment-14493633 ] 

Masood Parvari commented on DRILL-2765:
---------------------------------------

Mohsen is right. it is a very annoying bug. I wish you guys could release a patch as soon as possible instead of waiting till version 1.0
here I explain the problem with more details
Assume we are writing to parquet file using protobuf (or Avro). The proto file looks like this:

message Log  {
    optional string date = 1;
    optional string url = 2;
}

it is a reduced version of my problem. Now when writing to a parquet file (Non-hadoop), I want to have a case in which for one of the columns there exist no data at all (I am working on schema evolution). so I am just writing value for 'date' column in parquet file and column 'url' is completely null through all records. File is created successfully but when I am trying to query it by apache drill, It throws a null pointer exception. If at least for one of written records, I set 'url' field to a value, then it is ok and all other 'url' values can be null and drill can query it. But I need the case in which in a parquet file, a whole column is null (but other parquet files may have values for that column). please help me guys. my parquet version is : 1.6.0rc7 and apache drill version is 0.8.0.

> Throwing a  NPE on exceuting query against parquet files
> --------------------------------------------------------
>
>                 Key: DRILL-2765
>                 URL: https://issues.apache.org/jira/browse/DRILL-2765
>             Project: Apache Drill
>          Issue Type: Sub-task
>          Components: Storage - Parquet
>    Affects Versions: 0.8.0
>            Reporter: Mohsen Zainalpour
>            Assignee: Steven Phillips
>             Fix For: 1.0.0
>
>
> Executing a query on a parquet file that one of it`s columns haven`t any value at all(but field exists in the schema) thorws a NPE



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