[HIVE-4293] Predicates following UDTF operator are removed by PPD

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Fix Version/s: 0.13.0
• Component/s: Query Processor
• Assignee: Navis Ryu
• Reporter: Navis Ryu
• Created: 04/Apr/13 04:35
• Updated: 02/Jun/15 19:39
• Resolved: 22/Mar/14 05:04

Description

For example,

explain SELECT value from (
  select explode(array(key, value)) as (value) from (
    select * FROM src WHERE key > 200
  ) A
) B WHERE value > 300
;

Makes plan like this, removing last predicates

  TableScan
    alias: src
    Filter Operator
      predicate:
          expr: (key > 200.0)
          type: boolean
      Select Operator
        expressions:
              expr: array(key,value)
              type: array<string>
        outputColumnNames: _col0
        UDTF Operator
          function name: explode
          Select Operator
            expressions:
                  expr: col
                  type: string
            outputColumnNames: _col0
            File Output Operator
              compressed: false
              GlobalTableId: 0
              table:
                  input format: org.apache.hadoop.mapred.TextInputFormat
                  output format: org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat

Issue Links

• blocks: HIVE-4904 A little more CP crossing RS boundaries
• duplicates: HIVE-6395 multi-table insert from select transform fails if optimize.ppd enabled
• is blocked by: HIVE-4295 Lateral view makes invalid result if CP is disabled
• is blocked by: HIVE-4297 LvJ operator does not have colExprMap for columns from UDTF
• is duplicated by: HIVE-5964 Hive missing a filter predicate causing wrong results joining tables after sort by
• is related to: HIVE-10841 [WHERE col is not null] does not work sometimes for queries with many JOIN statements
• relates to: HIVE-4598 Incorrect results when using subquery in multi table insert
• links to: review board