lilei1128 opened a new pull request, #9922:
URL: https://github.com/apache/paimon/pull/9922
### Purpose
A bitmap file index can select rows that have already been deleted by
a deletion vector. In the data evolution single-file read path, the
index was evaluated without the DV, so the reader still opened the
data file even when every matching row was deleted. This change lets
the reader skip such files before opening them.
###Changes:
- Pass the group's DV to FileIndexEvaluator in the single-file path.
- Add an evaluator overload with a fileOffset, while preserving the
existing API with a default offset of zero.
- Convert anchor-relative DV positions to file-local positions before
intersecting the live rows with the file index result.
- Reuse the reader's DV offset calculation and range validation.
Keep reader-level DV filtering as the final correctness check.
### Tests
- Add FileIndexEvaluatorTest#testDataFilterIntersectsOffsetDeletionVector
to cover non-zero offset conversion and a neighboring deletion.
- Add
DataEvolutionFileIndexTest#testBitmapSelectionComposesWithDeletionVectorForNonZeroRowId
to verify that a deleted bitmap hit skips a missing data file in a
later row-id group while a live hit remains visible.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
