[jira] [Commented] (ORC-297) BloomFilterVersion.ORIGINAL is not honored

["dzcxzl (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/ORC-297?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17800800#comment-17800800 ] 

dzcxzl commented on ORC-297:
----------------------------

 Problems caused by Hive1.2's reconstruction of ORC reader([HIVE-9555|https://issues.apache.org/jira/browse/HIVE-9555]). 

BloomFilterUTF8 is read as a DATA stream.

https://github.com/apache/hive/blob/6002c510113d9a6aa87159c7386f2a8a4747405b/ql/src/java/org/apache/hadoop/hive/ql/io/orc/RecordReaderImpl.java#L939-L947

org.apache.hadoop.hive.ql.io.orc.RecordReaderImpl#createStreams
{code:java}
    for (OrcProto.Stream streamDesc: streamDescriptions) {
      int column = streamDesc.getColumn();
      if ((includeColumn != null && !includeColumn[column]) ||
              (streamDesc.hasKind() &&
          (StreamName.getArea(streamDesc.getKind()) != StreamName.Area.DATA))) {
        streamOffset += streamDesc.getLength();
        continue;
      }
{code}


> BloomFilterVersion.ORIGINAL is not honored
> ------------------------------------------
>
>                 Key: ORC-297
>                 URL: https://issues.apache.org/jira/browse/ORC-297
>             Project: ORC
>          Issue Type: Bug
>          Components: Java
>    Affects Versions: 1.4.2
>            Reporter: Seshu Pasam
>            Priority: Major
>
> When BloomFilterVersion.ORIGINAL is specified when creating ORC file, it creates both original bloom filter and the later UTF-8 version. This causes problem when third-party software like Hive (1.2), Presto (latest version) read the ORC. I can contribute patch/fix for this!



--
This message was sent by Atlassian Jira
(v8.20.10#820010)