You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
exmy
changed the title
[CH] Aligning the NULL and NaN sorting semantics of Spark and CH
[CH] Aligning the NULL and NaN compare semantics of Spark and CH
Jul 19, 2023
Spark and CH have different compare semantics for NULL and NaN value:
nan_direction_hint
can control its least or greatest.example:
sort array
[null,0.5,1.0,2.1,NaN]
in ascending, in descending[null,0.5,1.0,2.1,NaN]
[NaN,2.1,1.0,0.5,null]
[0.5,1.0,2.1,nan,NULL]
[2.1,1.0,0.5,nan,NULL]
The
array_max/array_min
functions have the same issue.It seems there is no other way to align them except by modifying the CH kernel code, and the modify is unlikely to be accepted by CH community.
The initial change:
The text was updated successfully, but these errors were encountered: