Skip to content

Incorrect filter pushdown in ORC when string stats truncated #2042

New issue
New issue
Open
Open
Incorrect filter pushdown in ORC when string stats truncated#2042
Labels
ORCcorrectnessa query provides different output from spark
@Flyangz

Description

@Flyangz
Flyangz
opened on Feb 27, 2026

Describe the bug
When string statistics are truncated (long strings), the predicate evaluation in row group filtering was incorrect.
datafusion-contrib/orc-rust#79

Additional context
Perhaps we should temporarily disable ORC predicate pushdown?

Metadata

Metadata

Assignees

No one assigned

    Labels

    ORCcorrectnessa query provides different output from spark

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions