haohuaijin opened a new issue, #8927: URL: https://github.com/apache/arrow-datafusion/issues/8927
### Describe the bug In our current implementation, the sub-expression within the short-circuited expression is always evaluated even when they don't need to be evaluated. We first evaluate all expressions recursively and then perform the corresponding operation or function. The related code is below the behavior of `COALESCE` function is wrong https://github.com/apache/arrow-datafusion/blob/b7e13a0af711477ad41450566c14430089edd3f2/datafusion/physical-expr/src/scalar_function.rs#L152-L156 the behavior of `And` and `Or` operation is wrong https://github.com/apache/arrow-datafusion/blob/b7e13a0af711477ad41450566c14430089edd3f2/datafusion/physical-expr/src/expressions/binary.rs#L262-L263 the behavior of `CASE ... WHEN ...` is correct https://github.com/apache/arrow-datafusion/blob/b7e13a0af711477ad41450566c14430089edd3f2/datafusion/physical-expr/src/expressions/case.rs#L136-L140 ### To Reproduce ``` DataFusion CLI v34.0.0 ❯ create table t(x int, y int) as values (1,1), (2,2), (3,3), (0,0), (4,0); 0 rows in set. Query took 0.004 seconds. ❯ select y > 0 and 1/y < 1 from t; Arrow error: Divide by zero error ❯ select y = 0 or 1/y < 1 from t; Arrow error: Divide by zero error ❯ select COALESCE(1, 1/y) from t; Arrow error: Divide by zero error ❯ select case 1 when 2 then 1/y end from t; +--------------------------------------------------------------------+ | CASE Int64(1) WHEN Int64(2) THEN Int64(1) / CAST(t.y AS Int64) END | +--------------------------------------------------------------------+ | | | | | | | | | | +--------------------------------------------------------------------+ 5 rows in set. Query took 0.003 seconds. ``` ### Expected behavior The short-circuited expressions should have the correct results. I also check in postgres ``` postgres=# create table t(x int, y int); insert into t values (1,1), (2,2), (3,3), (0,0), (4,0); CREATE TABLE INSERT 0 5 postgres=# select y > 0 and 1/y < 1 from t; ?column? ---------- f t t f f (5 rows) postgres=# select y = 0 or 1/y < 1 from t; ?column? ---------- f t t t t (5 rows) postgres=# select COALESCE(1, 1/y) from t; coalesce ---------- 1 1 1 1 1 (5 rows) ``` ### Additional context _No response_ -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
