[ https://issues.apache.org/jira/browse/PIG-1193?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12800984#action_12800984 ]
Daniel Dai commented on PIG-1193: --------------------------------- Diagnosis for this issue: {code} Reduce plan: Store(fakefile:org.apache.pig.builtin.PigStorage) - 1-56 | |---New For Each(false)[bag] - 1-55 | | | POUserFunc(sequence.CUMULATIVE)[bag] - 1-54 | | | |---RelationToExpressionProject[bag][*] - 1-49 | | | | | |---RelationToExpressionProject[bag][*] - 1-58 | | | | | |---Project[tuple][1] - 1-46 | | | |---RelationToExpressionProject[bag][*] - 1-53 | | | |---POSort[bag]() - 1-52 | | | | | Project[int][0] - 1-51 | | | |---Project[tuple][1] - 1-50 | |---Package[tuple]{chararray} - 1-43-------- {code} We take the first input's reverse POSort and make it a secondary sort key. However, we did not remove the second input's POSort. So the second input for the UDF is reverse reverse sorted. > Secondary sort issue on nested desc sort > ---------------------------------------- > > Key: PIG-1193 > URL: https://issues.apache.org/jira/browse/PIG-1193 > Project: Pig > Issue Type: Bug > Components: impl > Affects Versions: 0.6.0 > Reporter: Daniel Dai > Assignee: Daniel Dai > Fix For: 0.6.0 > > > Secondary sort doing nested desc sort order incorrectly if the following > conditions meet: > 1. We have sort and UDF in nested plan > 2. This UDF will use the same input tuples more than once > 3. The input tuples are sorted in desc order > Here is a test case: > {code} > register sequence.jar; > A = load 'input' as (a0:int); > B = group A ALL; > C = foreach B { > D = order A by a0 desc; > generate sequence.CUMULATIVE(D,D); > }; > dump C; > {code} > input file: > {code} > 3 > 4 > {code} > The input for the UDF is: > {code} > ({(4),(3)},{(3),(4)}) > {code} > The first bag is sorted desc, but the second is not. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.