[HIVE-6185] DDLTask is inconsistent in creating a table and adding a partition when dealing with location

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.12.0
• Fix Version/s: 0.13.0
• Component/s: Query Processor
• Assignee: Xuefu Zhang
• Reporter: Xuefu Zhang
• Created: 10/Jan/14 21:41
• Updated: 14/Jan/14 00:47
• Resolved: 14/Jan/14 00:47

Description

When creating a table, Hive uses URI to represent location:

    if (crtTbl.getLocation() != null) {
      tbl.setDataLocation(new Path(crtTbl.getLocation()).toUri());
    }

When adding a partition, Hive uses Path to represent location:

      // set partition path relative to table
      db.createPartition(tbl, addPartitionDesc.getPartSpec(), new Path(tbl
                    .getPath(), addPartitionDesc.getLocation()), addPartitionDesc.getPartParams(),
                    addPartitionDesc.getInputFormat(),
                    addPartitionDesc.getOutputFormat(),
                    addPartitionDesc.getNumBuckets(),
                    addPartitionDesc.getCols(),
                    addPartitionDesc.getSerializationLib(),
                    addPartitionDesc.getSerdeParams(),
                    addPartitionDesc.getBucketCols(),
                    addPartitionDesc.getSortCols());

This disparity makes the values stored in metastore be encoded differently, causing problems w.r.t. special character as demonstrated in HIVE-5446. As a result, the code dealing with location for table is different for partition, creating maintenance burden.

We need to standardize it to Path to be in line with other Path related cleanup effort.

Issue Links

• links to: https://reviews.apache.org/r/16806/