diff options
author | Reynold Xin <rxin@databricks.com> | 2015-07-25 01:28:46 -0700 |
---|---|---|
committer | Reynold Xin <rxin@databricks.com> | 2015-07-25 01:28:46 -0700 |
commit | f0ebab3f6d3a9231474acf20110db72c0fb51882 (patch) | |
tree | 6eeafa3d8066f970dbcb3cc85b80c726bea984a0 /unsafe | |
parent | 723db13e0688bf20e2a5f02ad170397c3a287712 (diff) | |
download | spark-f0ebab3f6d3a9231474acf20110db72c0fb51882.tar.gz spark-f0ebab3f6d3a9231474acf20110db72c0fb51882.tar.bz2 spark-f0ebab3f6d3a9231474acf20110db72c0fb51882.zip |
[SPARK-9336][SQL] Remove extra JoinedRows
They were added to improve performance (so JIT can inline the JoinedRow calls). However, we can also just improve it by projecting output out to UnsafeRow in Tungsten variant of the operators.
Author: Reynold Xin <rxin@databricks.com>
Closes #7659 from rxin/remove-joinedrows and squashes the following commits:
7510447 [Reynold Xin] [SPARK-9336][SQL] Remove extra JoinedRows
Diffstat (limited to 'unsafe')
0 files changed, 0 insertions, 0 deletions