You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
@Omega359 Thanks for the comment. Yes, this effort is also related with #14392.
I think we have 2 use cases for array_min & array_max functions (and i think this also common requirement for other SQL functions.). They are required for DataFusion end-users (not using Comet) and Comet support is another use-case because Spark supports them natively and DataFusion has not supported yet. We will also need for the mapping on Comet side for Spark queries using this scalar array functions submitted to DataFusion by Comet.
Activity
erenavsarogullari commentedon Feb 3, 2025
take
Omega359 commentedon Feb 3, 2025
Possibly related/tangential: #14392
erenavsarogullari commentedon Feb 3, 2025
@Omega359 Thanks for the comment. Yes, this effort is also related with #14392.
I think we have 2 use cases for
array_min&array_maxfunctions (and i think this also common requirement for other SQL functions.). They are required for DataFusion end-users (not using Comet) and Comet support is another use-case because Spark supports them natively and DataFusion has not supported yet. We will also need for the mapping on Comet side for Spark queries using this scalar array functions submitted to DataFusion by Comet.[-]Add array_min function support[/-][+]Add `array_min` function support[/+]Jefffrey commentedon Sep 27, 2025
Implemented by #16574
array_minscalar function #16570