You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@mahout.apache.org by "Trevor Grant (JIRA)" <ji...@apache.org> on 2017/04/22 18:38:04 UTC

[jira] [Created] (MAHOUT-1971) Aggregate Transpose Bug

Trevor Grant created MAHOUT-1971:
------------------------------------

             Summary: Aggregate Transpose Bug
                 Key: MAHOUT-1971
                 URL: https://issues.apache.org/jira/browse/MAHOUT-1971
             Project: Mahout
          Issue Type: Bug
            Reporter: Trevor Grant
            Assignee: Trevor Grant
             Fix For: 0.13.1


In the At operation of Flink and spark there is a line that looks like:

case (row, blockRow) => colV(row) = blockA(blockRow, blockCol)

that ought to be 

case (row, blockRow) => colV(row) += blockA(blockRow, blockCol)

Since it isnt'- the 'aggregate transpose' does not operate as expected. 

This needs to be updated in the Spark and Flink bindings (same one char fix). 

Test coverage should be updated to catch this.




--
This message was sent by Atlassian JIRA
(v6.3.15#6346)