You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@hive.apache.org by Szehon Ho <sz...@cloudera.com> on 2013/11/12 01:31:31 UTC

Review Request 15438: HIVE-5788 select * fails for table after adding new columns using rcfile storage format

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/15438/
-----------------------------------------------------------

Review request for hive.


Bugs: HIVE-5788
    https://issues.apache.org/jira/browse/HIVE-5788


Repository: hive-git


Description
-------

There was a subtle bug in the Rcfile's default serde : LazyBinaryColumnarSerde.  One of its member classes LazyBinaryColumnarStruct had an optimization to skip attempts to deserialize columns in a row that did not have a corresponding byte representation.  That is fine, but as the serde and its struct was getting re-used for reading subsequent rows, the flag was not being reset, and those columns would be marked for skipped even for later rows.


Diffs
-----

  serde/src/java/org/apache/hadoop/hive/serde2/columnar/ColumnarStructBase.java 0bc2c8f 
  serde/src/test/org/apache/hadoop/hive/serde2/columnar/TestLazyBinaryColumnarSerDe.java 99420ca 

Diff: https://reviews.apache.org/r/15438/diff/


Testing
-------

Manual testing, and adding a unit test for this serde.


Thanks,

Szehon Ho