You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@systemml.apache.org by "Matthias Boehm (JIRA)" <ji...@apache.org> on 2018/02/24 09:40:00 UTC
[jira] [Closed] (SYSTEMML-2075) Invalid constant folding of
matrix-scalar logical operations
[ https://issues.apache.org/jira/browse/SYSTEMML-2075?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Matthias Boehm closed SYSTEMML-2075.
------------------------------------
Resolution: Fixed
Assignee: Matthias Boehm
Fix Version/s: SystemML 1.1
> Invalid constant folding of matrix-scalar logical operations
> ------------------------------------------------------------
>
> Key: SYSTEMML-2075
> URL: https://issues.apache.org/jira/browse/SYSTEMML-2075
> Project: SystemML
> Issue Type: Bug
> Reporter: Matthias Boehm
> Assignee: Matthias Boehm
> Priority: Major
> Fix For: SystemML 1.1
>
>
> Matrix-scalar logical operations such as \{{C = A & FALSE}} are incorrectly rewritten to \{{FALSE}} during constant folding, which changes the output data type causing subsequent operations to fail. For example, \{{write(C, "tmp")}} crashes as follows:
>
> {code}
> PROGRAM ( size CP/SP = 0/1 )
> --MAIN PROGRAM
> ----GENERIC (lines 22-27) [recompile=true]
> ------SPARK write false.SCALAR.BOOLEAN.true target/testTemp/functions/binary/scalar/LogicalTest/out/C.SCALAR.STRING.true textcell.SCALAR.STRING.true .SCALAR.STRING.true
> Caused by: org.apache.sysml.runtime.DMLRuntimeException: Failed to obtain RDD for data type other than matrix or frame.
> at org.apache.sysml.runtime.controlprogram.context.SparkExecutionContext.getRDDHandleForVariable(SparkExecutionContext.java:325)
> at org.apache.sysml.runtime.instructions.spark.WriteSPInstruction.processFrameWriteInstruction(WriteSPInstruction.java:237)
> at org.apache.sysml.runtime.instructions.spark.WriteSPInstruction.processInstruction(WriteSPInstruction.java:145)
> at org.apache.sysml.runtime.controlprogram.ProgramBlock.executeSingleInstruction(ProgramBlock.java:264)
> {code}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)