linhongliu-db opened a new pull request, #59816: URL: https://github.com/apache/spark/pull/59816
### What changes were proposed in this pull request? Parenthesize NOT operands in inputToSQL, with a comment explaining why the existing child-count check misses this unary operator. ### Why are the changes needed? NOT has lower SQL precedence than comparisons. Ungrouped operands also change the IS NULL checks synthesized for null-safe equality. For example, `(NOT a) <=> b` with `a = true` and `b = false` should return true, but the ungrouped generated expression returns false. ### Does this PR introduce _any_ user-facing change? Yes, a bug fix. Connector comparison SQL retains the intended grouping of NOT, including the expansion of null-safe equality. ### How was this patch tested? Added SQL grouping assertions and 189 evaluations spanning all seven comparison operators, NOT on either or both sides, and every TRUE/FALSE/NULL input combination. The semantic regression fails on unchanged Apache code: (NOT a) <=> b with a=true and b=false returns false instead of true. On JDK 17, all 22 V2PredicateSuite tests and three JDBC filter/grouping tests passed. Catalyst Java Checkstyle, SQL test Scalastyle and git diff --check passed. ```bash build/sbt 'sql/testOnly org.apache.spark.sql.execution.datasources.v2.V2PredicateSuite' build/sbt 'sql/testOnly org.apache.spark.sql.jdbc.JDBCSuite -- -z "compile filters" -z "IS [NOT] NULL"' ``` ### Was this patch authored or co-authored using generative AI tooling? Generated-by: OpenAI Codex (GPT-6) -- 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]
