Skip to content

UDF fails when subexpression elimination is used in interpreted mode #803

@chris-twiner

Description

@chris-twiner

As UDF expects compilation always this fails when a child cannot compile, such as ExpressionProxy.

If, in interpreted mode, a child is deemed an identical subexpression (as occurs in the UdfTests."multiple two argument udf" test) spark is able to swap the contents as an optimisation to reduce re-calculation.

In the case where the analyser deems compilation the correct approach this isn't an issue. Targeted fix as part of #800.

Known to occur on Databricks 11.3 and other lower runtimes, fixed as part of 3.4 in SPARK-41991 .

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions