[ 
https://issues.apache.org/jira/browse/HADOOP-11505?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Chris Nauroth updated HADOOP-11505:
-----------------------------------
    Attachment: HADOOP-11505.007.patch

AFAIK, the only time you'll see big endian on Windows is the XBox.  I'm not 
much of a gamer these days, but last time I was in a game shop, I didn't see 
Hadoop on the shelf next to Fallout and Halo, so I'm going to make the bold 
claim that we can always assume little endian for our Windows build.  :-)

The perfect way to solve this would be to convert the Windows build to use 
CMake like everything else.  HADOOP-11080 tracks that effort, but it's going to 
take a long time.  I wouldn't want to block this patch waiting for HADOOP-11080 
to complete.

In the interim, we can have the native-win profile approximate this part of the 
CMake script by running a few Ant tasks to generate hadoop_endian.h with the 
correct macros for Windows.  I'm attaching patch v007 with this solution.  
Compared to the prior patch, the only new changes are in pom.xml and 
native.vcxproj, so it should be a small incremental code review.

I think this is slightly better than checking in a hard-coded hadoop_endian.h 
just for Windows.  It's possible this way will give us less of a 
dual-maintenance burden.


> Various native parts use bswap incorrectly and unportably
> ---------------------------------------------------------
>
>                 Key: HADOOP-11505
>                 URL: https://issues.apache.org/jira/browse/HADOOP-11505
>             Project: Hadoop Common
>          Issue Type: Bug
>    Affects Versions: 3.0.0
>            Reporter: Colin Patrick McCabe
>            Assignee: Alan Burlison
>             Fix For: 3.0.0
>
>         Attachments: HADOOP-11505.001.patch, HADOOP-11505.003.patch, 
> HADOOP-11505.004.patch, HADOOP-11505.005.patch, HADOOP-11505.006.patch, 
> HADOOP-11505.007.patch
>
>
> hadoop-mapreduce-client-nativetask fails to use x86 optimizations in some 
> cases.  Also, on some alternate, non-x86, non-ARM architectures the generated 
> code is incorrect.  Thanks to Steve Loughran and Edward Nevill for finding 
> this.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to