[
https://issues.apache.org/jira/browse/PIG-1695?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12926495#action_12926495
]
Thejas M Nair commented on PIG-1695:
------------------------------------
+1
With this fix, pig does not merge the foreach statements if first foreach has a
user defined schema. We should consider supporting that case as well (in a
separate jira), because users sometimes add a foreach statement just to give
convenient names for columns .
eg-
{code}
J = join A by col1, B by col1
F1 = foreach J generate A::col1 as Acol1 , A::col2 as col2, B::col1 as Bcol1;
-- foreach that has been added just to give convenient names for expressions.
F2 = foreach F1 generate Acol1+col2, Bcol1 + col2 ;
{code}
> MergeForEach does not carry user defined schema if any one of the merged
> ForEach has user defined schema
> --------------------------------------------------------------------------------------------------------
>
> Key: PIG-1695
> URL: https://issues.apache.org/jira/browse/PIG-1695
> Project: Pig
> Issue Type: Bug
> Components: impl
> Affects Versions: 0.8.0
> Reporter: Daniel Dai
> Assignee: Daniel Dai
> Fix For: 0.8.0
>
> Attachments: PIG-1695-1.patch
>
>
> The following script missing the column name i after merge.
> {code}
> a = load 'num.txt' as (i);
> b = foreach a generate (int)i;
> c = foreach b generate i + 60 as i;
> store c into 'sectest';
> {code}
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
