Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
Hive, mail # dev - Review Request 20399: Invalid column access info for partitioned table


Copy link to this message
-
Re: Review Request 20399: Invalid column access info for partitioned table
Ashutosh Chauhan 2014-05-12, 20:21

This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/20399/#review42627
Patch looks good. But looks like there are few changes which may not be essential for the patch.
ql/src/java/org/apache/hadoop/hive/ql/exec/TableScanOperator.java
<https://reviews.apache.org/r/20399/#comment76596>

    Its not clear whats the difference between neededColumns & referencedColumns. If not, can we just use neededColumns? If there is any, it would be good to add a comment, why neededColumns is not sufficient here.

ql/src/java/org/apache/hadoop/hive/ql/exec/TableScanOperator.java
<https://reviews.apache.org/r/20399/#comment76454>

    Operator should not contain any compile time info, only runtime info. Compile time info belongs to Desc classes. So, move this field to TableScanDesc class.

ql/src/java/org/apache/hadoop/hive/ql/optimizer/ColumnPrunerProcFactory.java
<https://reviews.apache.org/r/20399/#comment76455>

    In line with above comment, this should then be scanOp.getConf().setReferencedColumns()

ql/src/java/org/apache/hadoop/hive/ql/optimizer/ppr/PartitionPruner.java
<https://reviews.apache.org/r/20399/#comment76598>

    Its not clear how this referredColumns are used. Its populated, but seems like no one is making use of it.
- Ashutosh Chauhan
On May 7, 2014, 4:06 a.m., Navis Ryu wrote: