[jira] [Commented] (HIVE-5802) Hive UDF pow limits the second input to integer when the first input is a decimal

["Xuefu Zhang (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HIVE-5802?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13837072#comment-13837072 ] 

Xuefu Zhang commented on HIVE-5802:
-----------------------------------

This is fixed in HIVE-5706. Here is the new console output:
{code}
hive> explain select pow(dd, 2) from test limit 1;  
OK
ABSTRACT SYNTAX TREE:
  (TOK_QUERY (TOK_FROM (TOK_TABREF (TOK_TABNAME test1))) (TOK_INSERT (TOK_DESTINATION (TOK_DIR TOK_TMP_FILE)) (TOK_SELECT (TOK_SELEXPR (TOK_FUNCTION pow (TOK_TABLE_OR_COL dd) 2))) (TOK_LIMIT 1)))

STAGE DEPENDENCIES:
  Stage-1 is a root stage
  Stage-0 is a root stage

STAGE PLANS:
  Stage: Stage-1
    Map Reduce
      Alias -> Map Operator Tree:
        test1 
          TableScan
            alias: test1
            Select Operator
              expressions:
                    expr: power(dd, 2)
                    type: double
              outputColumnNames: _col0
              Limit
                File Output Operator
                  compressed: false
                  GlobalTableId: 0
                  table:
                      input format: org.apache.hadoop.mapred.TextInputFormat
                      output format: org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat
                      serde: org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe

  Stage: Stage-0
    Fetch Operator
      limit: 1
{code}

> Hive UDF pow limits the second input to integer when the first input is a decimal
> ---------------------------------------------------------------------------------
>
>                 Key: HIVE-5802
>                 URL: https://issues.apache.org/jira/browse/HIVE-5802
>             Project: Hive
>          Issue Type: Improvement
>          Components: Types, UDF
>    Affects Versions: 0.12.0
>            Reporter: Xuefu Zhang
>            Assignee: Xuefu Zhang
>
> For example,
> {code}
> hive> desc test;
> OK
> i                   	int                 	None                
> b                   	boolean             	None                
> d                   	double              	None                
> s                   	string              	None                
> dec                 	decimal(5,2)        	None                
> hive> explain select pow(dec, 2.5) from test;
> FAILED: SemanticException [Error 10014]: Line 1:15 Wrong arguments '2.5': No matching method for class org.apache.hadoop.hive.ql.udf.UDFPower with (decimal(5,2), double). Possible choices: _FUNC_(decimal(65,30), int)  _FUNC_(double, double)  _FUNC_(double, int)  
> {code}
> This seems too restrictive. MySQL, on the other hand, supports fractional power as the second input.
> {code}
> mysql> select pow(d, 2.5) from test;
> mysql> desc test;
> +-------+--------------+------+-----+---------+-------+
> | Field | Type         | Null | Key | Default | Extra |
> +-------+--------------+------+-----+---------+-------+
> | i     | int(11)      | YES  |     | NULL    |       |
> | b     | tinyint(1)   | YES  |     | NULL    |       |
> | d     | double       | YES  |     | NULL    |       |
> | s     | varchar(5)   | YES  |     | NULL    |       |
> | dd    | decimal(5,2) | YES  |     | NULL    |       |
> +-------+--------------+------+-----+---------+-------+
> 5 rows in set (0.00 sec)
> mysql> select pow(dd, 2.5) from test;
> +-------------------+
> | pow(dd, 2.5)      |
> +-------------------+
> | 2570.215713318881 |
> +-------------------+
> 1 row in set (0.00 sec)
> {code}
> Hive should also support this.



--
This message was sent by Atlassian JIRA
(v6.1#6144)