feat: add the similar optimization function for bitwise negative #5516
Conversation
| /// ~(A | B) ===> ~A & ~B | ||
| /// ~(~A) ===> A | ||
| /// For others, use Negative clause | ||
| pub fn negative_clause(expr: Expr) -> Expr { |
There was a problem hiding this comment.
I think, it deserves a better name.
There was a problem hiding this comment.
I think this is basically applying https://en.wikipedia.org/wiki/De_Morgan%27s_laws
Maybe we can call it distribute_negation?
|
Regarding tests: I have not noticed relevant tests for this segment of the code: https://github.com/apache/arrow-datafusion/blob/main/datafusion/optimizer/src/simplify_expressions/utils.rs#L249-#L273. Should we create a separate PR for this case to keep the symmetry? |
Yes please. Or even better I wonder if there is a way to avoid the duplication. Thank you @izveigor |
| /// ~(A | B) ===> ~A & ~B | ||
| /// ~(~A) ===> A | ||
| /// For others, use Negative clause | ||
| pub fn negative_clause(expr: Expr) -> Expr { |
There was a problem hiding this comment.
I think this is basically applying https://en.wikipedia.org/wiki/De_Morgan%27s_laws
Maybe we can call it distribute_negation?
|
Converted to draft as it is waiting on other PRs and tests |
3 participants
Which issue does this PR close?
Follow on to #5423
Rationale for this change
I propose to add the symmetric optimization function for bitwise negative.
Rules for bitwise negative in case of a binary expression are the same as the not clause:
~(A & B) ===> ~A | ~B
~(A | B) ===> ~A & ~B
~(~A) ===> A
What changes are included in this PR?
New optimization method for binary expressions with bitwise negative.
Are these changes tested?
After PR merge: #5511
Are there any user-facing changes?
Should increase execution speed