Repository navigation
feat: Introduce ansi enable parameter for execution config #18634
Description
Is your feature request related to a problem or challenge?
As was discussed with @Omega359 and @Jefffrey in apache/datafusion-spark#5 (comment) :
A question I have is how do we plan to support spark.sql.ansi.enabled config? I've seen a few PRs for Spark functions that try to cater for this config, but we don't exactly have a way to set this so it usually results in dead code (e.g. a hardcoded boolean that is set to false for now) or their own way of toggling it (e.g. another argument to the function).
ConfigOptions via ScalarFunctionArgs.
This ANSI flag becomes more critical as Spark 4.0 configured the ANSI mode as true by default
https://spark.apache.org/docs/latest/sql-ref-ansi-compliance.html
Describe the solution you'd like
No response
Describe alternatives you've considered
No response
Additional context
No response
Reacted by hsiang-c, Martin, Bruce Ritchie and Bhargava Vadlamani
Activity
[-]feat: Introduce `ansi` enabled parameter for execution config[/-][+]feat: Introduce `ansi` enable parameter for execution config[/+]hsiang-c commentedon Nov 11, 2025
take (if no one else is working on it)
comphead commentedon Nov 11, 2025
Sorry @hsiang-c the PR is already assigned #18635
feat: Add `ansi` enable parameter for execution config (#18635)
coderfender commentedon Nov 29, 2025
@comphead, I believe we need to make
eval_modeas an ENUM rather than a boolean given how the behavior between Try / Legacy mode is not always the same. For example, an overflow in SUM results in wrap around value in Legacy mode while it is null inTry. I can create an issue and work on it to convert the bool to an ENUM if thats okay