diff options
author | Cheng Lian <lian@databricks.com> | 2017-03-06 10:36:50 -0800 |
---|---|---|
committer | Cheng Lian <lian@databricks.com> | 2017-03-06 10:36:50 -0800 |
commit | 339b53a1311e08521d84a83c94201fcf3c766fb2 (patch) | |
tree | 929a7dc8b83acd79676049e9b0a43b771a54f7ed /sql/core/src/main | |
parent | 2a0bc867a4a1dad4ecac47701199e540d345ff4f (diff) | |
download | spark-339b53a1311e08521d84a83c94201fcf3c766fb2.tar.gz spark-339b53a1311e08521d84a83c94201fcf3c766fb2.tar.bz2 spark-339b53a1311e08521d84a83c94201fcf3c766fb2.zip |
[SPARK-19737][SQL] New analysis rule for reporting unregistered functions without relying on relation resolution
## What changes were proposed in this pull request?
This PR adds a new `Once` analysis rule batch consists of a single analysis rule `LookupFunctions` that performs simple existence check over `UnresolvedFunctions` without actually resolving them.
The benefit of this rule is that it doesn't require function arguments to be resolved first and therefore doesn't rely on relation resolution, which may incur potentially expensive partition/schema discovery cost.
Please refer to [SPARK-19737][1] for more details about the motivation.
## How was this patch tested?
New test case added in `AnalysisErrorSuite`.
[1]: https://issues.apache.org/jira/browse/SPARK-19737
Author: Cheng Lian <lian@databricks.com>
Closes #17168 from liancheng/spark-19737-lookup-functions.
Diffstat (limited to 'sql/core/src/main')
0 files changed, 0 insertions, 0 deletions