Describe the bug
row_number and group by have inconsistent partition results for (0.0, -0.0)
To Reproduce
A.k1 contains -0.0 and 0.0
The result of select k1, count(1) from A group by k1 is as follows.
+------+----------+
| k1 | count(1) |
+------+----------+
| 0 | 2 |
| -0 | 1 |
+------+----------+
The result of select k1, row_number() over (partition by k1 order by k2) as t from A is as follows.
+------+------+
| k1 | t |
+------+------+
| 0 | 1 |
| 0 | 2 |
| -0 | 3 |
+------+------+
Expected behavior
We hope to unify the partition results of group by and row_number