sunchao opened a new pull request, #11135:
URL: https://github.com/apache/arrow-rs/pull/11135
# Which issue does this PR close?
Follow-up regression coverage for #10733 and #10735; does not close an
additional issue.
# Rationale for this change
The regression added by #10735 starts with an explicit sparse row selection
and one predicate. This test covers sparse selection produced by the first
predicate, followed by a second predicate on a different column, using the
default Auto policy and predicate-cache capacity. It complements the broader
single-predicate matrix in #10823.
# What changes are included in this PR?
Add one async-reader test with three 20-row pages and an 8-row batch size.
The key predicate selects rows 0 and 40, skipping the middle page. The value
predicate and output projection exercise cached reads of the remaining column.
Assert the exact output values and that records were read from the cache.
# Are these changes tested?
- `cargo +stable fmt --all -- --check` passes.
- `rustfmt +stable --check --edition 2024
parquet/tests/arrow_reader/row_filter/async.rs` passes.
- `git diff --check` passes.
- Execution of the focused test is pending. The command below is blocked
during dependency resolution, before compilation: this environment's dependency
cooldown does not yet provide `crc32fast 1.5.2`, required by the unchanged
upstream lockfile. Parquet tests and Clippy have therefore not been validated
locally.
```shell
cargo +stable test --locked -p parquet --features async --test arrow_reader \
'row_filter::r#async::test_default_auto_filter_cache_with_skipped_page' \
-- --exact --nocapture
```
# Are there any user-facing changes?
No. This change adds one test and does not modify runtime behavior or public
APIs.
# AI assistance disclosure
Codex adapted the existing regression to the current page-index API and
prepared the validation notes and PR description. The dependency-resolution
limitation above remains outstanding.
--
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]
