peterxcli opened a new pull request, #5194:
URL: https://github.com/apache/datafusion-comet/pull/5194
## Which issue does this PR close?
Follow-up to #5176. Related to #5191.
## Rationale for this change
PR #5176 was merged only after I filed #5191, but the changes requested in
its final review had not been pushed yet. This follow-up publishes those review
changes on top of the merged upstream `main`.
## What changes are included in this PR?
- Build one nested comparator over the unsliced child arrays and reuse it
for every row, using absolute offsets.
- Preserve left/right comparator argument order when the smaller right-hand
array becomes the probe side.
- Add a regression test for sliced arrays with non-zero offsets and the
probe-side swap.
- Add short and long nested-list and nested-struct benchmarks.
- Reference #5191 from the signed-zero tests so the known Spark mismatch is
not presented as intended behavior.
## How are these changes tested?
- `cargo test -p datafusion-comet-spark-expr --lib` (590 passed)
- `cargo clippy -p datafusion-comet-spark-expr --lib --tests --benches -- -D
warnings`
- `cargo fmt --all -- --check`
- Criterion before/after:
- nested int32 short: 128.41 ms -> 1.818 ms (-98.586%)
- nested int32 long: 133.93 ms -> 1.657 ms (-98.767%)
- nested struct short: 68.97 ms -> 1.041 ms (-98.466%)
- nested struct long: 65.88 ms -> 0.963 ms (-98.541%)
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
