Compare commits

...

4 Commits

Author SHA1 Message Date
dependabot[bot]
195e69dac4 Bump actions/checkout from 6.0.3 to 7.0.0
Bumps [actions/checkout](https://github.com/actions/checkout) from 6.0.3 to 7.0.0.
- [Release notes](https://github.com/actions/checkout/releases)
- [Changelog](https://github.com/actions/checkout/blob/main/CHANGELOG.md)
- [Commits](df4cb1c069...9c091bb21b)

---
updated-dependencies:
- dependency-name: actions/checkout
  dependency-version: 7.0.0
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>
2026-06-19 20:03:50 +00:00
Pascal Seitz
1e859fd78d fix term aggregation u32::MAX overflow issue 2026-06-18 17:07:43 +08:00
Pascal Seitz
f451fa938f explain why naive scorer must accumulate scores in WAND order 2026-06-17 18:58:58 +08:00
Pascal Seitz
2a82dd6f64 fix flaky test 2026-06-17 18:58:58 +08:00
9 changed files with 227 additions and 11 deletions

View File

@@ -20,7 +20,7 @@ jobs:
contents: read
steps:
- uses: actions/checkout@df4cb1c069e1874edd31b4311f1884172cec0e10 # v6.0.3
- uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
- name: Install Rust
run: rustup toolchain install nightly-2025-12-01 --profile minimal --component llvm-tools-preview
- uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1

View File

@@ -25,7 +25,7 @@ jobs:
contents: read
steps:
- uses: actions/checkout@df4cb1c069e1874edd31b4311f1884172cec0e10 # v6.0.3
- uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
- name: Install stable
uses: actions-rs/toolchain@16499b5e05bf2e26879000db0c1d13f7e13fa3af # v1.0.7
with:

View File

@@ -22,7 +22,7 @@ jobs:
steps:
- name: 'Checkout code'
uses: actions/checkout@df4cb1c069e1874edd31b4311f1884172cec0e10 # v6.0.3
uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
with:
persist-credentials: false

View File

@@ -27,7 +27,7 @@ jobs:
checks: write
steps:
- uses: actions/checkout@df4cb1c069e1874edd31b4311f1884172cec0e10 # v6.0.3
- uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
- name: Install nightly
uses: actions-rs/toolchain@16499b5e05bf2e26879000db0c1d13f7e13fa3af # v1.0.7
@@ -77,7 +77,7 @@ jobs:
name: test-${{ matrix.features.label}}
steps:
- uses: actions/checkout@df4cb1c069e1874edd31b4311f1884172cec0e10 # v6.0.3
- uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
- name: Install stable
uses: actions-rs/toolchain@16499b5e05bf2e26879000db0c1d13f7e13fa3af # v1.0.7

View File

@@ -275,7 +275,7 @@ impl SegmentCompositeCollector {
dict.insert(
key,
IntermediateCompositeBucketEntry {
doc_count: agg.count,
doc_count: agg.count as u64,
sub_aggregation: sub_aggregation_res,
},
);

View File

@@ -957,7 +957,7 @@ fn into_intermediate_bucket_entry(
)?;
}
Ok(IntermediateTermBucketEntry {
doc_count: bucket.count,
doc_count: bucket.count as u64,
sub_aggregation: sub_aggregation_res,
})
}

View File

@@ -98,7 +98,7 @@ impl SegmentAggregationCollector for TermMissingAgg {
let missing_count = &self.missing_count_per_bucket[parent_bucket_id as usize];
let mut missing_entry = IntermediateTermBucketEntry {
doc_count: missing_count.missing_count,
doc_count: missing_count.missing_count as u64,
sub_aggregation: Default::default(),
};
if let Some(sub_agg) = &mut self.sub_agg {

View File

@@ -930,7 +930,7 @@ impl IntermediateRangeBucketEntry {
#[derive(Clone, Default, Debug, PartialEq, Serialize, Deserialize)]
pub struct IntermediateTermBucketEntry {
/// The number of documents in the bucket.
pub doc_count: u32,
pub doc_count: u64,
/// The sub_aggregation in this bucket.
pub sub_aggregation: IntermediateAggregationResults,
}
@@ -1240,6 +1240,24 @@ mod tests {
assert_eq!(tree_left, tree_expected);
}
#[test]
fn test_term_bucket_doc_count_no_u32_overflow() {
// Two segments each contributing (u32::MAX - 100) docs to the same term. Summing them
// overflowed when doc_count was u32.
let per_segment = u32::MAX as u64 - 100;
let mut entry = IntermediateTermBucketEntry {
doc_count: per_segment,
sub_aggregation: Default::default(),
};
entry
.merge_fruits(IntermediateTermBucketEntry {
doc_count: per_segment,
sub_aggregation: Default::default(),
})
.unwrap();
assert_eq!(entry.doc_count, per_segment * 2);
}
#[test]
fn test_merge_fruits_tree_empty() {
let mut tree_left = get_intermediate_tree_with_ranges(&[

View File

@@ -273,8 +273,14 @@ mod tests {
}
if all_match {
let score: Score =
leader.score() + secondaries.iter_mut().map(|s| s.score()).sum::<Score>();
// Accumulate in the same left-to-right order as the WAND implementation
// (leader first, then each secondary in turn). Float addition is not
// associative, so `leader + secondaries.sum()` gives a different bit
// pattern and can cause spurious nearly_equals failures.
let mut score: Score = leader.score();
for secondary in secondaries.iter_mut() {
score += secondary.score();
}
if score > limit {
heap.push(Float(score));
@@ -417,6 +423,198 @@ mod tests {
}
}
#[test]
fn test_block_wand_intersection_three_scorers_regression() {
// Minimal failing case found by proptest (CI run 27557430583, job 81460063906).
// Posting list 0 spans docs 063 (all present, doc 8 has tf=80, doc 26 tf=4, rest tf=1).
// Posting lists 1 and 2 are sparse with varying term freqs, and doc 16/64 appear only
// in lists 1/2 but not list 0. The high tf=80 on doc 8 of list 0 makes the WAND
// upper-bound estimation skip documents that the naive intersection would score.
let posting_lists: &[&[(DocId, u32)]] = &[
&[
(0, 1),
(1, 1),
(2, 1),
(3, 1),
(4, 1),
(5, 1),
(6, 1),
(7, 1),
(8, 80),
(9, 1),
(10, 1),
(11, 1),
(12, 1),
(13, 1),
(14, 1),
(15, 1),
(17, 1),
(18, 1),
(19, 1),
(20, 1),
(21, 1),
(22, 1),
(23, 1),
(24, 1),
(25, 1),
(26, 4),
(27, 1),
(28, 1),
(29, 1),
(30, 1),
(31, 1),
(32, 1),
(33, 1),
(34, 1),
(35, 1),
(36, 1),
(37, 1),
(38, 1),
(39, 1),
(40, 1),
(41, 1),
(42, 1),
(43, 1),
(44, 1),
(45, 1),
(46, 1),
(47, 1),
(48, 1),
(49, 1),
(50, 1),
(51, 1),
(52, 1),
(53, 1),
(54, 1),
(55, 1),
(56, 1),
(57, 1),
(58, 1),
(59, 1),
(60, 1),
(61, 1),
(62, 1),
(63, 1),
],
&[
(0, 2),
(3, 98),
(7, 93),
(8, 87),
(9, 39),
(10, 2),
(12, 71),
(14, 47),
(15, 76),
(16, 6),
(17, 38),
(19, 61),
(20, 87),
(21, 1),
(22, 5),
(23, 43),
(25, 48),
(26, 87),
(28, 81),
(29, 69),
(30, 7),
(31, 47),
(32, 32),
(33, 38),
(35, 39),
(38, 65),
(39, 98),
(42, 43),
(43, 52),
(44, 99),
(45, 88),
(48, 24),
(51, 61),
(52, 22),
(53, 58),
(55, 26),
(56, 32),
(58, 57),
(60, 29),
(61, 78),
(62, 9),
(63, 44),
(64, 29),
],
&[
(0, 94),
(2, 49),
(3, 63),
(4, 7),
(6, 93),
(7, 17),
(8, 91),
(9, 18),
(10, 85),
(11, 11),
(12, 45),
(13, 42),
(15, 91),
(16, 44),
(17, 36),
(18, 68),
(19, 24),
(20, 17),
(21, 59),
(22, 97),
(24, 20),
(25, 7),
(26, 85),
(27, 69),
(28, 78),
(29, 84),
(30, 35),
(31, 49),
(33, 83),
(34, 97),
(35, 29),
(36, 43),
(37, 59),
(38, 79),
(39, 74),
(40, 21),
(41, 5),
(42, 47),
(43, 27),
(44, 59),
(45, 97),
(46, 91),
(47, 81),
(48, 57),
(49, 47),
(50, 64),
(51, 86),
(52, 60),
(53, 52),
(54, 14),
(55, 23),
(56, 64),
(57, 40),
(58, 5),
(59, 30),
(60, 81),
(61, 62),
(62, 39),
(63, 93),
(64, 82),
],
];
let fieldnorms: &[u32] = &[
624, 668, 725, 670, 851, 169, 537, 627, 200, 757, 51, 272, 835, 89, 750, 63, 272, 406,
394, 390, 822, 449, 257, 571, 527, 855, 4, 98, 548, 413, 539, 351, 596, 151, 728, 152,
766, 829, 20, 828, 477, 251, 743, 646, 136, 477, 909, 907, 266, 341, 676, 161, 40, 384,
347, 707, 42, 397, 482, 814, 801, 528, 465, 410, 171,
];
let posting_lists_owned: Vec<Vec<(DocId, u32)>> =
posting_lists.iter().map(|pl| pl.to_vec()).collect();
test_block_wand_intersection_aux(&posting_lists_owned, fieldnorms);
}
#[test]
fn test_block_wand_intersection_disjoint() {
// Two posting lists with no overlap — intersection is empty.