Github user heary-cao commented on the issue:
https://github.com/apache/spark/pull/18961
@gatorsmile , This should be a problem for code execution, semantics, and
consistency.
This trigger condition: _FileSourceScanExec.partitionFilters_ is not null
and contains nondeterministic function. however, the current spark master
branch does not trigger this code path. because _PhysicalOperation_ excluded
that the condition is nondeterministic of Filter.
Similar spark sql: ` SELECT t1.i3 from tableorc t1 where rand(10) <= 0.5`
the current spark master branch executed Plan:
```
*Project [i3#0]
+- *Filter (rand(10) <= 0.5)
+- *FileScan orc
default.tableorc[i3#0,i4#1,i5#2,i6#3,i7#4,i8#5,i9#6,i10#7,i11#8,s2#9,s3#10,s4#11,s5#12,s6#13,s7#14,d2#15,d3#16,d4#17,d5#18,d6#19,d7#20,i2#21]
Batched: false, Format: ORC, Location:
CatalogFileIndex[file:/home/cxw/spark/bin/spark-warehouse/tableorc],
PartitionCount: 2, PartitionFilters: [], PushedFilters: [], ReadSchema:
struct<i3:int,i4:int,i5:int,i6:int,i7:int,i8:int,i9:int,i10:int,i11:int,s2:string,s3:string,s4:st...
```
PartitionFilters is **[]**
Thus, this trigger condition is associated with #18918. it will be more
prone to this exception.
#18918 PR executed Plan:
```
*Project [i3#0]
+- *Filter (rand(10) <= 0.5)
+- *FileScan orc default.tableorc[i3#0,i2#21] Batched: false, Format:
ORC, Location:
PrunedInMemoryFileIndex[file:/home/cxw/spark/bin/spark-warehouse/tableorc/i2=0],
PartitionCount: 1, PartitionFilters: [(rand(10) <= 0.5)], PushedFilters: [],
ReadSchema: struct<i3:int>
```
PartitionFilters is **[(rand(10) <= 0.5)]**
thanks.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
