You are viewing a plain text version of this content. The canonical link for it is here.
Posted to github@datafusion.apache.org by "Nagato-Yuzuru (via GitHub)" <gi...@apache.org> on 2026/10/06 15:06:51 UTC
[PR] test: generate NULLs in equivalence fuzz tests [datafusion]
Nagato-Yuzuru opened a new pull request, #26086:
URL: https://github.com/apache/datafusion/pull/26086
## Which issue does this PR close?
<!--
We generally require a GitHub issue to be filed for all bug fixes and enhancements and this helps us generate change logs for our releases. You can link an issue to this PR using the GitHub syntax. For example `Closes #123` indicates that this PR will close issue #123.
-->
Part of #24243
## Rationale for this change
<!--
Why are you proposing this change? If this is already explained clearly in the issue then this section is not needed.
Explaining clearly why changes are proposed helps reviewers understand your changes and offer better suggestions for fixes.
Please explain the problem you are trying to solve in terms of the user-visible
behavior, rather than the implementation.
For example, "The code in `foo.rs` doesn't handle nulls" is a symptom of the
implementation. "COUNT(DISTINCT) returns wrong results when the column contains
nulls" is the user-visible problem.
-->
The equivalence fuzz tests never generated NULLs. So they could not catch NULL-ordering bugs in `SortProperties` propagation such as #11596 and #24208.
## What changes are included in this PR?
<!--
There is no need to duplicate the description in the issue here, but it is sometimes worth providing a summary of the individual changes in this PR.
-->
- Generated columns contain NULLs at configurable rates (`NULL_PCTS = [0.0, 0.1, 0.5]`), crossed with every seed; the schema is nullable only when the rate is non-zero.
- `create_random_schema` picks random `SortOptions` for each ordering key.
- Requirements with random `SortOptions` are checked only for soundness: if `ordering_satisfy` accepts an ordering, the data must be sorted by it. The converse is not checked, since one batch can be sorted by coincidence (e.g. NULLs of independent columns landing in the same rows). For the same reason, `contains_overflowable_arithmetic` is renamed to `contains_conservative_ordering_op` and also covers comparisons and `AND`/`OR`.
## What is the testing strategy for this PR?
<!--
We typically require tests for all PRs in order to:
1. Prevent the code from being accidentally broken by subsequent changes
2. Serve as another way to document the expected behavior of the code
Briefly describe how this PR is tested, and point to the specific tests you added. For example: 'This new feature is covered by the `sqllogictest` cases added in `foo.slt`'.
If this PR does not add tests, explain why. For example, if the change is already covered by existing tests, please mention it.
You should also check the `codecov` bot reply on this PR to confirm the changed code is exercised.
-->
Test-only change. Verified locally that reverting the fixes from #24276 and #24206 makes `test_ordering_satisfy_with_equivalence_complex_random` fail, and that it passes on main:
```bash
git revert -n ceaef16778f3422312cc1e22ddf40e8f6acda56b 4042812a86f4a753c4b717ecbe12dbc928957254
cargo test -p datafusion --test fuzz --features extended_tests equivalence
# test_ordering_satisfy_with_equivalence_complex_random fails
```
## Are there any user-facing changes?
<!--
If there are user-facing changes then we may require documentation to be updated before approving the PR.
If there are any breaking changes to public APIs, please add the `api change` label.
-->
No
--
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: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] test: generate NULLs in equivalence fuzz tests [datafusion]
Posted by "alamb (via GitHub)" <gi...@apache.org>.
alamb commented on PR #26086:
URL: https://github.com/apache/datafusion/pull/26086#issuecomment-6382875195
This PR appears to be waiting on the author to respond to reviewer feedback so marking as draft. Please mark ready to review when it is ready for another look
--
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: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] test: generate NULLs in equivalence fuzz tests [datafusion]
Posted by "kosiew (via GitHub)" <gi...@apache.org>.
kosiew commented on code in PR #26086:
URL: https://github.com/apache/datafusion/pull/26086#discussion_r4347057350
##########
datafusion/core/tests/fuzz_cases/equivalence/ordering.rs:
##########
@@ -127,10 +171,21 @@ fn test_ordering_satisfy_with_equivalence_complex_random() -> Result<()> {
col("f", &test_schema)?,
floor_a,
a_plus_b,
+ a_gt_b,
+ a_gt_e_and_b_gt_e,
+ a_gt_e_or_b_gt_e,
];
+ let mut options_rng = StdRng::seed_from_u64(seed as u64);
for n_req in 1..=exprs.len() {
for exprs in exprs.iter().combinations(n_req) {
Review Comment:
Optional follow-up: could we measure this loop under the CI profile? The expansion from 8 to 11 expressions and three NULL rates increases fixed-option cases from 25,500 to 614,100 (about 24x), plus one random-option inference check per case. This is not a measured runtime increase. If CI cost warrants it, consider retaining exhaustive coverage of the original expressions while testing singleton/pair combinations of new expressions and deterministically sampling longer combinations. Please verify that historical NULL-ordering regressions remain detectable.
##########
datafusion/core/tests/fuzz_cases/equivalence/utils.rs:
##########
@@ -503,16 +561,25 @@ pub fn generate_table_for_orderings(
Ok(batch)
}
+pub const NULL_PCTS: &[f64] = &[0.0, 0.1, 0.5];
Review Comment:
Optional follow-up: could we add deterministic cases for an all-NULL ordered column and an explicitly NULL constant operand, with matching schema nullability and both sort directions/NULL placements? The current 0%, 10%, and 50% matrix does not guarantee these boundaries, and generated constant columns remain non-NULL. Use the one-way soundness assertion rather than adding 100% NULLs to bidirectional checks, and explicitly assert expected singleton inference where guaranteed so rejection cannot make the test vacuous.
--
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: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] test: generate NULLs in equivalence fuzz tests [datafusion]
Posted by "Nagato-Yuzuru (via GitHub)" <gi...@apache.org>.
Nagato-Yuzuru commented on code in PR #26086:
URL: https://github.com/apache/datafusion/pull/26086#discussion_r4375817826
##########
datafusion/core/tests/fuzz_cases/equivalence/utils.rs:
##########
@@ -503,16 +561,25 @@ pub fn generate_table_for_orderings(
Ok(batch)
}
+pub const NULL_PCTS: &[f64] = &[0.0, 0.1, 0.5];
Review Comment:
Added both cases in f49799e15. Without the Singleton guard in SortProperties::and/or, the NULL constant test fails. An all-NULL column cannot break an ordering, so that test fails only when an expected ordering is not inferred.
--
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: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] test: generate NULLs in equivalence fuzz tests [datafusion]
Posted by "codecov-commenter (via GitHub)" <gi...@apache.org>.
codecov-commenter commented on PR #26086:
URL: https://github.com/apache/datafusion/pull/26086#issuecomment-6020041089
## [Codecov](https://app.codecov.io/gh/apache/datafusion/pull/26086?dropdown=coverage&src=pr&el=h1&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=apache) Report
:white_check_mark: All modified and coverable lines are covered by tests.
:white_check_mark: Project coverage is 82.72%. Comparing base ([`31a4ca0`](https://app.codecov.io/gh/apache/datafusion/commit/31a4ca07fe1f874a5dec3dc79f4e53b8107a5a20?dropdown=coverage&el=desc&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=apache)) to head ([`4c5cdc6`](https://app.codecov.io/gh/apache/datafusion/commit/4c5cdc6594a58f5ffe008af7aafccae89511d35d?dropdown=coverage&el=desc&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=apache)).
:warning: Report is 214 commits behind head on main.
<details><summary>Additional details and impacted files</summary>
```diff
@@ Coverage Diff @@
## main #26086 +/- ##
==========================================
+ Coverage 82.42% 82.72% +0.29%
==========================================
Files 1139 1147 +8
Lines 435357 448179 +12822
Branches 435357 448179 +12822
==========================================
+ Hits 358844 370740 +11896
- Misses 54812 54902 +90
- Partials 21701 22537 +836
```
</details>
[:umbrella: View full report in Codecov by Harness](https://app.codecov.io/gh/apache/datafusion/pull/26086?dropdown=coverage&src=pr&el=continue&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=apache).
:loudspeaker: Have feedback on the report? [Share it here](https://about.codecov.io/codecov-pr-comment-feedback/?utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=apache).
<details><summary> :rocket: New features to boost your workflow: </summary>
- :snowflake: [Test Analytics](https://docs.codecov.com/docs/test-analytics): Detect flaky tests, report on failures, and find test suite problems.
- :package: [JS Bundle Analysis](https://docs.codecov.com/docs/javascript-bundle-analysis): Save yourself from yourself by tracking and limiting bundle sizes in JS merges.
</details>
--
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: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] test: generate NULLs in equivalence fuzz tests [datafusion]
Posted by "Nagato-Yuzuru (via GitHub)" <gi...@apache.org>.
Nagato-Yuzuru commented on code in PR #26086:
URL: https://github.com/apache/datafusion/pull/26086#discussion_r4375789276
##########
datafusion/core/tests/fuzz_cases/equivalence/ordering.rs:
##########
@@ -127,10 +171,21 @@ fn test_ordering_satisfy_with_equivalence_complex_random() -> Result<()> {
col("f", &test_schema)?,
floor_a,
a_plus_b,
+ a_gt_b,
+ a_gt_e_and_b_gt_e,
+ a_gt_e_or_b_gt_e,
];
+ let mut options_rng = StdRng::seed_from_u64(seed as u64);
for n_req in 1..=exprs.len() {
for exprs in exprs.iter().combinations(n_req) {
Review Comment:
Thanks. I measured it locally with --profile ci --features extended_tests:
| | before this PR | this PR, no sampling | this PR, sampling |
|-------------------------------------------------------|----------------|----------------------|-------------------|
| test_ordering_satisfy_with_equivalence_complex_random | 2.0 s | 82.9 s | 11.5 s |
To check that NULL ordering regressions stay detectable, I removed the NULL guards in SortProperties::and/or and in SortProperties::gt_or_gteq locally, one at a time. The sampled test fails in both cases.
--
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: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org