GitHub user HyukjinKwon opened a pull request:
https://github.com/apache/spark/pull/21882
[SPARK-24934][SQL] Handle missing upper/lower bounds case in in-memory
partition pruning
## What changes were proposed in this pull request?
Looks we intentionally set `null` for lower bounds for complex types and
don't use it. However, these look used in in-memory partition pruning, which
ends up with incorrect results.
This PR proposes to don't filter on `null` when both bounds are `null`.
This can be false-positive but still better than not working.
```scala
val df = Seq(Array("a", "b"), Array("c", "d")).toDF("arrayCol")
df.cache().filter("arrayCol > array('a', 'b')").show()
```
**Before:**
```
Predicate isnotnull(arrayCol#3) generates partition filter:
((arrayCol.count#18 - arrayCol.nullCount#17) > 0)
Predicate (arrayCol#3 > [a,b]) generates partition filter: ([a,b] <
arrayCol.upperBound#15)
Skipping partition based on stats arrayCol.lowerBound: null,
arrayCol.upperBound: null, arrayCol.nullCount: 0, arrayCol.count: 1,
arrayCol.sizeInBytes: 52
+--------+
|arrayCol|
+--------+
+--------+
```
**After:**
```
Predicate isnotnull(arrayCol#3) generates partition filter:
((arrayCol.count#18 - arrayCol.nullCount#17) > 0)
Predicate (arrayCol#3 > [a,b]) generates partition filter:
(isnull(arrayCol.upperBound#15) || ([a,b] < arrayCol.upperBound#15))
+--------+
|arrayCol|
+--------+
| [c, d]|
+--------+
```
## How was this patch tested?
Unit tests were added and manually tested.
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/HyukjinKwon/spark stats-filter
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/spark/pull/21882.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #21882
----
commit ea38b56dcdd3010c70fcfbddd4405917278431cd
Author: hyukjinkwon <gurwls223@...>
Date: 2018-07-26T12:34:44Z
Handle missing upper/lower bounds case in inmemory partition pruning
----
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
