Fix RowConverter when FixedSizeList is not the last #7789
Merged
Conversation
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
Fix `RowConverter` row decoding when there is a `FixedSizeList` element and it's not the last.
findepi
commented
Jun 26, 2025
|
🤖 |
|
🤖: Benchmark completed Details
|
When columns are unequal sizes, then encoding them can panic. This can be observed by changing `test_fixed_size_list_with_variable_width_content`, eg adding one more element to the second list. What's unclear, the panic can be avoided by calling `tracker.materialized()` in `compute_lengths_fixed_size_list` (similar to how list encoding does it). However, we still won't encode the whole data passed in, so it's better to reject this upfront.
findepi
commented
Jun 27, 2025
Comment on lines
+693
to
+701
| for colum in columns.iter().skip(1) { | ||
| if colum.len() != columns[0].len() { | ||
| return Err(ArrowError::InvalidArgumentError(format!( | ||
| "RowConverter columns must all have the same length, expected {} got {}", | ||
| columns[0].len(), | ||
| colum.len() | ||
| ))); | ||
| } | ||
| } |
There was a problem hiding this comment.
Another check added. This guards against passing ill-formed data from the user.
Without this, it's possible to get a panic inside the RowEncoder.
|
🤖 |
|
I am not sure what happened to the benchmark script -- I will rerun shortly |
|
🤖 |
|
🤖: Benchmark completed Details
|
|
I will rerun the benchmarks on this PR again -- some of the results look like noise |
|
🤖 |
|
🤖: Benchmark completed Details
|
12 tasks
|
It looks like |
Possibly -- it could also be noise in the benchmark the next step woudl be to see if we can reproduce the slowdown locally (run |
|
The starting point of this PR, aka the merge base of this PR, is 01c5efc. and then (still on the same commit) Some benchmarks returned 19% improvement, some 19% or even 46% degradation. Definitely running on my laptop can hardly be called a well isolated environment, but i did my best in this regard. third run transcript |
alamb
reviewed
Jul 3, 2025
arrow-row/src/lib.rs
Outdated
| unsafe { self.convert_raw(&mut rows, validate_utf8) } | ||
| let result = unsafe { self.convert_raw(&mut rows, validate_utf8) }?; | ||
|
|
||
| for (i, row) in rows.iter().enumerate() { |
There was a problem hiding this comment.
Suggested change
| for (i, row) in rows.iter().enumerate() { | |
| #[cfg!(test)] | |
| for (i, row) in rows.iter().enumerate() { |
There was a problem hiding this comment.
I find if cfg!(test) { easier to read. Would it be ok?
Out of concern this can cause runtime penalty.
alamb
approved these changes
Jul 3, 2025
findepi
deleted the
findepi/fix-rowconverter-when-fixedsizelist-is-not-the-last-b962b2
branch
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Which issue does this PR close?
none
Rationale for this change
RowConverterdecoding fails when there is aFixedSizeListelement and it's not the last.What changes are included in this PR?
Fix
RowConverterrow decoding when there is aFixedSizeListelement and it's not the last.Are these changes tested?
yes