diff options
author | Reynold Xin <rxin@databricks.com> | 2015-07-30 13:09:43 -0700 |
---|---|---|
committer | Reynold Xin <rxin@databricks.com> | 2015-07-30 13:09:43 -0700 |
commit | a20e743fb863de809863652931bc982aac2d1f86 (patch) | |
tree | f948f807fb34b8861809e91a8b04fa22fb752af9 /ec2 | |
parent | 6d94bf6ac10ac851636c62439f8f2737f3526a2a (diff) | |
download | spark-a20e743fb863de809863652931bc982aac2d1f86.tar.gz spark-a20e743fb863de809863652931bc982aac2d1f86.tar.bz2 spark-a20e743fb863de809863652931bc982aac2d1f86.zip |
[SPARK-9460] Fix prefix generation for UTF8String.
Previously we could be getting garbage data if the number of bytes is 0, or on JVMs that are 4 byte aligned, or when compressedoops is on.
Author: Reynold Xin <rxin@databricks.com>
Closes #7789 from rxin/utf8string and squashes the following commits:
86ffa3e [Reynold Xin] Mask out data outside of valid range.
4d647ed [Reynold Xin] Mask out data.
c6e8794 [Reynold Xin] [SPARK-9460] Fix prefix generation for UTF8String.
Diffstat (limited to 'ec2')
0 files changed, 0 insertions, 0 deletions