Hi all,

I found a potential violation of the iceberg scan planning spec for
position deletes with referenced data files:
https://iceberg.apache.org/spec/#scan-planning

The spec states that a position delete applies to a data file when all of
the following are true:
- The delete file's partition spec and partition value match the data file's
- The data file's partition (both spec and partition values) is equal to
the delete file's partition

However, scan planning ignores partition mismatches between data and delete
files, associating the delete with the data file based solely on the path
reference. This appears to contradict the "all of the following"
requirement in the spec.

See the linked PR for a test case to reproduce:
https://github.com/apache/iceberg/pull/16939

Would appreciate feedback to ensure I am interpreting the spec correctly
before I propose a fix.

Thanks,
Grant

Reply via email to