Dandandan commented on PR #19732: URL: https://github.com/apache/datafusion/pull/19732#issuecomment-3734098320
> CI found an error using the sqlite test suite: > > ``` > External error: 1 errors in file /__w/datafusion/datafusion/datafusion-testing/data/sqlite/random/expr/slt_good_102.slt > > 1. query is expected to fail, but actually succeed: > [SQL] SELECT 57 * - - 12 * + + 83 + - 74 + - + CASE + ( COUNT ( * ) ) WHEN + 60 THEN + CAST ( NULL AS INTEGER ) ELSE NULL END + COUNT ( ALL - 4 + 61 / 0 ) * 29 * 28 > at /__w/datafusion/datafusion/datafusion-testing/data/sqlite/random/expr/slt_good_102.slt:8867 > ``` > > Makes sense that we should not remove case expressions that could cause runtime-failures (divide by zero error in this case). > > The question is how to detect if an expression could cause a side effect. I couldn't find any `Expr::side_effect_free` (or `Expr::has_side_effect`). Am I missing something? Does anyone have ideas on how this could be handled? We could of course restrict this optimization to only apply to simpler expressions like literals or columns, but it would be cool if we could apply this optimization more generally I think we could mark certain operators as pure/safe (will never cause runtime error) and operators as division as non-pure and only optimize subtrees with only pure / non-failing expressions. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
