spark-instrumented-optimizer/sql/core
Xingbo Jiang 031910b0ec [SPARK-21608][SPARK-9221][SQL] Window rangeBetween() API should allow literal boundary
## What changes were proposed in this pull request?

Window rangeBetween() API should allow literal boundary, that means, the window range frame can calculate frame of double/date/timestamp.

Example of the use case can be:
```
SELECT
	val_timestamp,
	cate,
	avg(val_timestamp) OVER(PARTITION BY cate ORDER BY val_timestamp RANGE BETWEEN CURRENT ROW AND interval 23 days 4 hours FOLLOWING)
FROM testData
```

This PR refactors the Window `rangeBetween` and `rowsBetween` API, while the legacy user code should still be valid.

## How was this patch tested?

Add new test cases both in `DataFrameWindowFunctionsSuite` and in `window.sql`.

Author: Xingbo Jiang <xingbo.jiang@databricks.com>

Closes #18814 from jiangxb1987/literal-boundary.
2017-08-09 13:23:49 +08:00
..
benchmarks [SPARK-17335][SQL] Fix ArrayType and MapType CatalogString. 2016-09-03 19:02:20 +02:00
src [SPARK-21608][SPARK-9221][SQL] Window rangeBetween() API should allow literal boundary 2017-08-09 13:23:49 +08:00
pom.xml [SPARK-20433][BUILD] Bump jackson from 2.6.5 to 2.6.7.1 2017-08-08 18:15:29 -07:00