[
https://issues.apache.org/jira/browse/PIG-2290?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13106899#comment-13106899
]
Thejas M Nair commented on PIG-2290:
------------------------------------
I didn't notice that the output does not match schema. That means one can't use
the output of the TOBAG later in the query. So fixing this is not likely to
have an impact on most of the existing queries. It will affect only some cases
where the output of TOBAG is stored without further processing on it into a
storage that does not store schema.
But this use case is something that users would need to check for, when the
upgrade. I would vote for making this change only in 0.10, and not for 0.9.1 .
> TOBAG wraps tuple parameters in another tuple
> ---------------------------------------------
>
> Key: PIG-2290
> URL: https://issues.apache.org/jira/browse/PIG-2290
> Project: Pig
> Issue Type: Bug
> Components: internal-udfs
> Affects Versions: 0.9.0
> Reporter: Ryan Hoegg
> Assignee: Dmitriy V. Ryaboy
> Attachments: pig-2290.patch
>
>
> The TOBAG function indiscriminately wraps all parameters in a tuple. When I
> pass a list of tuples to the function, I would expect it to return a bag
> containing those tuples. Instead, it returns a bag containing single element
> tuples, where each tuple contains one of the tuples passed in.
> Example:
> {code:title=tuples.txt}
> (mike,608)
> (ryan,11624)
> (justin,2317)
> {code}
> {code:title=Demonstration using pig 0.9.0}
> grunt> TUPLE_DATA = LOAD 'tuples.txt' AS
> (T:tuple(name:chararray,street_number:int));
> grunt> BAGGED = FOREACH TUPLE_DATA GENERATE TOBAG(T);
> grunt> DESCRIBE BAGGED;
> BAGGED: {{(name: chararray,street_number: int)}}
> grunt> DUMP BAGGED;
> ({((mike,608))})
> ({((ryan,11624))})
> ({((justin,2317))})
> {code}
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
