Bikramjeet Singh Vig

Fix handling of common sub expressions and encoded input

2

This patch fixes the following: - Handling common sub expressions running over disjoint set of active rows - Tracking of rows that have been pre-computed by the common expressions -...

Fix loss of precision in timestamp when spilling

8

Currently we use a presto compatible serializer when spilling which serializes timestamp to millisecond precision. This results in loss of precision for velox timestamp type since velox stores it in...

Fix NaN handling for array_remove and contains UDFs

2

Summary: Ensures that NaNs are considered as being equal to each other. Differential Revision: D57342173

Fix handling of NaN for map variants

7

Summary: Highlights of the this change: - Introduces a utility functors for floating point types that provide comparator and hash functor to implement consistent behavior or NaNs across the codebase....

Fix NaN handling for multiple array UDFs

3

Summary: Fixes for the following: array_sort array_distinct array_intersect array_except array_overlap array_union Differential Revision: D57305880

Fix NaN handling for array_min and array_max

6

Summary: Ensure that: - Null is returned consistently for all types - NaN is treated as greater than infinity Differential Revision: D57237545

Fix NaN handling for multimap_agg

7

Summary: Ensure multimap_agg treats all NaN keys as equal. Differential Revision: D57217450

Inconsistent Null handling in array_max/min functions in presence of NaN

2

presto returns a NaN, even if there is a NULL in the array. However, for other types, it always returns NULL if one exists. As a part of fixing the...

Add ability to verify expression fuzzer runs on a subset of rows

2

Summary: Currently, the expression fuzzer has a phase where it re-runs rows that did not throw an error to ensure evaluation is consistent for them. To achieve this, it currently...