fix(parquet): fix CDC panic on nested ListArrays with null entries #9644
Merged
alamb merged 4 commits intoapache:mainfrom Apr 7, 2026
Merged
fix(parquet): fix CDC panic on nested ListArrays with null entries #9644alamb merged 4 commits intoapache:mainfrom
alamb merged 4 commits intoapache:mainfrom
Conversation
…pache#9637) The CDC chunker's value_offset diverged from actual leaf array positions when null list entries had non-empty child offset ranges (valid per the Arrow columnar format spec). This caused slice_for_chunk to produce incorrect non_null_indices, leading to an out-of-bounds panic in write_mini_batch. Track non-null value counts (nni) separately from leaf slot counts in the chunker, and use them in slice_for_chunk to correctly index into non_null_indices regardless of gaps in the leaf array.
Member
Author
Contributor
|
🔨 🐛 |
alamb
pushed a commit
that referenced
this pull request
Apr 10, 2026
# Which issue does this PR close? N/A # Rationale for this change The content defined chunking bench (CDC) was disabled due to a bug (#9637). That has since been fixed (#9644), so the bench can be re-enabled. # What changes are included in this PR? Uncomments out the relevant bench in arrow_writer # Are these changes tested? N/A # Are there any user-facing changes? No
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The CDC chunker's value_offset diverged from actual leaf array positions when null list entries had non-empty child offset ranges (valid per the Arrow columnar format spec). This caused slice_for_chunk to produce incorrect non_null_indices, leading to an out-of-bounds panic in write_mini_batch.
Track non-null value counts (nni) separately from leaf slot counts in the chunker, and use them in slice_for_chunk to correctly index into non_null_indices regardless of gaps in the leaf array.
Which issue does this PR close?
Closes #9637
Rationale for this change
The Arrow spec allows null list entries to own non-empty child segments. When such arrays were written with CDC enabled,
CdcChunk::value_offsetdiverged from the actual index intonon_null_indices, causing an out-of-bounds panic.What changes are included in this PR?
CdcChunk::value_offset/num_valuesas index/count intonon_null_indicesinstead of leaf array positions.leaf_offsetin the nested branch to track leaf array position for hashing separately fromvalue_offset.slice_for_chunkto directly indexnon_null_indices.Are these changes tested?
Yes. Unit tests for
slice_for_chunk(nested nulls, all-null chunks) and end-to-end roundtrip tests for lists with non-empty null segments.Are there any user-facing changes?
No.