audit dict | base_commit stringlengths 40 40 | distribution dict | instance_id stringlengths 21 24 | packaging dict | pr_number int64 2.37k 6.7k | pr_url stringlengths 44 47 | problem_statement dict | reference_solution_patch dict | repo stringclasses 2
values | schema_version stringclasses 1
value | source dict | stage stringclasses 1
value | test_patch dict | verification dict |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "28ebf2fde5421b3c73a60127f3630fdf05032c773490b01f5b437641e967574b",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "a1493b9e07... | 0ca4c6109cdbb561576025be323814be4261c95d | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": true,
"has_tests_changed": true,
"language_area": "mixed",
"language_areas": [
"rust",
"python",
"format_spec"
],
"module_area": "dataset_table",
"module_areas": [
"dataset_table",
"versioning_tran... | lance-format__lance-3863 | {
"combined_selected": true,
"combined_selection_rank": 1,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 1,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir": "... | 3,863 | https://github.com/lance-format/lance/pull/3863 | {
"excluded_sources": [
"labels",
"tests"
],
"source_strategy": "linked_issue",
"text": "When using the Lance Python API, performing a `merge_insert` with a *partial schema* (i.e., the incoming table does not include all columns) can tombstone columns in existing fragments and write new column data. If ... | {
"bytes": 26664,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-3863.source.patch",
"private_artifact": true,
"sha256": "0dce0bfc638d51e195bfcd431bfad9ba636834279f56a8c1b0be54943fa68756",
"source_path": "data/lance-format__lance/ben... | lance-format/lance | benchmark-instance-v1 | {
"additions": 319,
"author_association": "MEMBER",
"base_commit": "0ca4c6109cdbb561576025be323814be4261c95d",
"base_ref": "main",
"changed_file_paths": [
"protos/transaction.proto",
"python/python/lance/dataset.py",
"python/python/tests/test_scalar_index.py",
"python/src/transaction.rs",
... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 1299,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-3863.test.patch",
"sha256": "c572afee0b47070d18181a83311348ea529eae146c252515521e351b7fdceb79",
"source_path": "data/lance-format__lance/benchmark_package... | {
"base_commit": "0ca4c6109cdbb561576025be323814be4261c95d",
"commands": [
{
"command": "cd python && /opt/venv/bin/python -m maturin develop --extras tests && /opt/venv/bin/python -m pytest -vvv -s -m 'not recurring and not integration and not slow' python/tests/test_scalar_index.py::test_index_after_mer... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "6f9bf3a0ae18370bc872e8705000130c7ba0972b50e8ff693954ca6fb27fefd8",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "032f050ead... | 677d44a5d1e4afed6c5481aee74a5d93c50da25c | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": false,
"has_tests_changed": false,
"language_area": "rust",
"language_areas": [
"rust"
],
"module_area": "dataset_table",
"module_areas": [
"dataset_table"
],
"patch_size_bucket": "small",
"problem_sourc... | lance-format__lance-3896 | {
"combined_selected": true,
"combined_selection_rank": 2,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 2,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir": "... | 3,896 | https://github.com/lance-format/lance/pull/3896 | {
"excluded_sources": [
"linked_issues",
"tests"
],
"source_strategy": "title_body",
"text": "Lance dataset metadata includes a manifest that may be cached for reuse across operations. A bug occurs when working with datasets created or upgraded across a manifest naming-scheme migration: cache lookup / r... | {
"bytes": 618,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-3896.source.patch",
"private_artifact": true,
"sha256": "f7c96857d791c6d0d551336d24280cdef521241541a3649af9c4c5974a24c625",
"source_path": "data/lance-format__lance/bench... | lance-format/lance | benchmark-instance-v1 | {
"additions": 22,
"author_association": "CONTRIBUTOR",
"base_commit": "f295e144facdc7796a69f4fadad16b1feb02c453",
"base_ref": "main",
"changed_file_paths": [
"rust/lance/src/dataset.rs"
],
"changed_files": 1,
"ci_check_suite_count": 7,
"ci_conclusion": "success",
"ci_failed_check_suite_count": ... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 1035,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-3896.test.patch",
"sha256": "a2be2327201876881312efd93794b8b85d5aceda7dcce455067a1e458caa7cef",
"source_path": "data/lance-format__lance/benchmark_package... | {
"base_commit": "677d44a5d1e4afed6c5481aee74a5d93c50da25c",
"commands": [
{
"command": "RUSTC_BOOTSTRAP=1 RUSTFLAGS='-Zcrate-attr=recursion_limit=\"256\"' cargo +stable test -p lance test_migrate_v2_manifest_paths",
"command_inline_test_names": [
"test_migrate_v2_manifest_paths"
],
... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "5c6068600a376b31ae5298b2674ea00bd1ce2b37bd57ff117f9e8f7d6730d347",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "10a84b6265... | 8b96f1b54550406939486dcd11bcd43d238aa5cf | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": false,
"has_tests_changed": false,
"language_area": "rust",
"language_areas": [
"rust"
],
"module_area": "query_scan",
"module_areas": [
"query_scan"
],
"patch_size_bucket": "small",
"problem_source": "t... | lance-format__lance-3937 | {
"combined_selected": true,
"combined_selection_rank": 3,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 3,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir": "... | 3,937 | https://github.com/lance-format/lance/pull/3937 | {
"excluded_sources": [
"linked_issue",
"tests"
],
"source_strategy": "title_body",
"text": "When Lance encodes a scalar expression (e.g., a filter predicate or computed expression) into Substrait, the resulting encoded representation may omit or incorrectly specify the schema context for that expressio... | {
"bytes": 3906,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-3937.source.patch",
"private_artifact": true,
"sha256": "f705d35d1d25f501f24e9d21c748db2378796a2872bc793a03287ce640b60732",
"source_path": "data/lance-format__lance/benc... | lance-format/lance | benchmark-instance-v1 | {
"additions": 39,
"author_association": "MEMBER",
"base_commit": "8b96f1b54550406939486dcd11bcd43d238aa5cf",
"base_ref": "main",
"changed_file_paths": [
"rust/lance-datafusion/src/substrait.rs"
],
"changed_files": 1,
"ci_check_suite_count": 6,
"ci_conclusion": "success",
"ci_failed_check_suite_... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 981,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-3937.test.patch",
"sha256": "12a14c9863bdbd1f5caf370199d333ffd105a8b08115f7499aa0b7a832b1f96c",
"source_path": "data/lance-format__lance/benchmark_package_... | {
"base_commit": "8b96f1b54550406939486dcd11bcd43d238aa5cf",
"commands": [
{
"command": "cargo +stable test -p lance-datafusion --features substrait test_expr_substrait_roundtrip",
"command_inline_test_names": [
"test_expr_substrait_roundtrip"
],
"command_mode": "per_inline_test_... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "cfae238207c538993eeea18ca1111071ca40487788134559363e17416c8e5469",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "aad2d3f5c0... | 177cbaaedad659073794bcc66ee8f5d65d4e2e92 | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": false,
"has_tests_changed": true,
"language_area": "mixed",
"language_areas": [
"rust",
"python"
],
"module_area": "dataset_table",
"module_areas": [
"dataset_table",
"python_api"
],
"patch_size_bu... | lance-format__lance-4001 | {
"combined_selected": true,
"combined_selection_rank": 4,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 4,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir": "... | 4,001 | https://github.com/lance-format/lance/pull/4001 | {
"excluded_sources": [
"linked_issues",
"tests"
],
"source_strategy": "pr_body",
"text": "When listing tags for a Lance dataset, the returned order is not guaranteed to be by version, which makes it harder to find the most recent tag.\n\nUpdate the tag-listing behavior so that tags are returned sorted ... | {
"bytes": 12018,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-4001.source.patch",
"private_artifact": true,
"sha256": "fa5fe7ab00516e1a902301766978866c2005e204c9979f0ce5fd26308aa0f3b6",
"source_path": "data/lance-format__lance/ben... | lance-format/lance | benchmark-instance-v1 | {
"additions": 219,
"author_association": "CONTRIBUTOR",
"base_commit": "201a94d902d98fb7a86491ffb16dc41dc525cc22",
"base_ref": "main",
"changed_file_paths": [
"docs/tags.rst",
"python/python/lance/dataset.py",
"python/python/lance/lance/__init__.pyi",
"python/python/tests/test_dataset.py",
... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 1688,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-4001.test.patch",
"sha256": "bf388da166da525b13d7d06176cb58abc9b559f64b48a4e4105eed8b9717dbe6",
"source_path": "data/lance-format__lance/benchmark_package... | {
"base_commit": "177cbaaedad659073794bcc66ee8f5d65d4e2e92",
"commands": [
{
"command": "cd python && /opt/venv/bin/python -m maturin develop --extras tests && /opt/venv/bin/python -m pytest -vvv -s -m 'not recurring and not integration and not slow' python/tests/test_dataset.py::test_tag_order",
"c... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "4a47887b966a88440a66dba9c88406e30702163e6ec0b7e4726cf2bae9147635",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "7757627141... | e7851863c7abb65d0b35708940215761a0f85608 | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": false,
"has_tests_changed": false,
"language_area": "rust",
"language_areas": [
"rust"
],
"module_area": "file_format_encoding",
"module_areas": [
"file_format_encoding"
],
"patch_size_bucket": "medium",
... | lance-format__lance-4157 | {
"combined_selected": true,
"combined_selection_rank": 5,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 5,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir": "... | 4,157 | https://github.com/lance-format/lance/pull/4157 | {
"excluded_sources": [
"linked_issue",
"tests"
],
"source_strategy": "pr_body",
"text": "When reading Lance data files, the file reader currently does not provide a public way for users to control caching behavior for the “full zip repetition index” used during reads. This makes it difficult to tune me... | {
"bytes": 14091,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-4157.source.patch",
"private_artifact": true,
"sha256": "e88647fbfc60e4b8b4846dfbb026292d299b6a6a07cfa47c3e8a15f579f7ea2c",
"source_path": "data/lance-format__lance/ben... | lance-format/lance | benchmark-instance-v1 | {
"additions": 160,
"author_association": "COLLABORATOR",
"base_commit": "e7851863c7abb65d0b35708940215761a0f85608",
"base_ref": "main",
"changed_file_paths": [
"rust/lance-encoding/src/decoder.rs",
"rust/lance-encoding/src/encodings/logical/primitive.rs",
"rust/lance-encoding/src/testing.rs",
... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 4346,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-4157.test.patch",
"sha256": "589627b9c90d44b74f3b955768e5bfa3e641412adca1ba2b1ef3d85b91fed860",
"source_path": "data/lance-format__lance/benchmark_package... | {
"base_commit": "e7851863c7abb65d0b35708940215761a0f85608",
"commands": [
{
"command": "cargo +stable test -p lance-encoding test_fullzip_cache_config_controls_caching",
"command_inline_test_names": [
"test_fullzip_cache_config_controls_caching"
],
"command_mode": "per_inline_te... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "b9de2383f96ff169be6920431958277eb16eb2bb16efa3ec74500f442effc36d",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "e09150af6f... | 7f489db354b7017d84c768b933d48d69611d2192 | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": true,
"has_tests_changed": false,
"language_area": "rust",
"language_areas": [
"rust"
],
"module_area": "build_packaging",
"module_areas": [
"build_packaging"
],
"patch_size_bucket": "small",
"problem_so... | lance-format__lance-4226 | {
"combined_selected": true,
"combined_selection_rank": 6,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 6,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir": "... | 4,226 | https://github.com/lance-format/lance/pull/4226 | {
"excluded_sources": [
"tests"
],
"source_strategy": "linked_issue",
"text": "When applying the `normalize_fsl` operation to an Arrow array that contains null values, the result does not properly respect the input null mask. This can lead to outputs where values corresponding to input nulls are treated as ... | {
"bytes": 2380,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-4226.source.patch",
"private_artifact": true,
"sha256": "729c45d51316a06aeb5c51cf382296eda930ab643b80b38df20d3949768ebdd5",
"source_path": "data/lance-format__lance/benc... | lance-format/lance | benchmark-instance-v1 | {
"additions": 92,
"author_association": "COLLABORATOR",
"base_commit": "7f489db354b7017d84c768b933d48d69611d2192",
"base_ref": "main",
"changed_file_paths": [
"Cargo.lock",
"python/Cargo.lock",
"rust/lance-linalg/Cargo.toml",
"rust/lance-linalg/src/kernels.rs"
],
"changed_files": 4,
"ci... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 3507,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-4226.test.patch",
"sha256": "7158a8524e0277c887b76dc1510e67cf745319e7d7f26e89a318ec51d45505e0",
"source_path": "data/lance-format__lance/benchmark_package... | {
"base_commit": "7f489db354b7017d84c768b933d48d69611d2192",
"commands": [
{
"command": "cargo +stable test -p lance-linalg test_normalize_fsl_with_nulls",
"command_inline_test_names": [
"test_normalize_fsl_with_nulls"
],
"command_mode": "per_inline_test_name",
"language": ... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "022fa264be900832cd4046fa3847508babfcd86c2ea30697211c460ea4137b96",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "491f30776e... | 6e987921d0efbe42c018047ea45d1b6f624d8280 | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": false,
"has_tests_changed": true,
"language_area": "rust",
"language_areas": [
"rust"
],
"module_area": "index_search",
"module_areas": [
"index_search"
],
"patch_size_bucket": "small",
"problem_source":... | lance-format__lance-4235 | {
"combined_selected": true,
"combined_selection_rank": 7,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 7,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir": "... | 4,235 | https://github.com/lance-format/lance/pull/4235 | {
"excluded_sources": [
"linked_issues (none)",
"pr_body (empty)",
"tests (not provided)"
],
"source_strategy": "title_body",
"text": "Lance currently supports “prewarming” some index types to reduce first-query latency by loading the index into memory ahead of time. However, B-Tree (scalar) indices... | {
"bytes": 1878,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-4235.source.patch",
"private_artifact": true,
"sha256": "eff4e84255a175c428103ddf9f06d5a0dd5a78defc871ef81ccd3bf1e5edbfd3",
"source_path": "data/lance-format__lance/benc... | lance-format/lance | benchmark-instance-v1 | {
"additions": 52,
"author_association": "CONTRIBUTOR",
"base_commit": "6e987921d0efbe42c018047ea45d1b6f624d8280",
"base_ref": "main",
"changed_file_paths": [
"python/python/tests/test_scalar_index.py",
"rust/lance-index/src/scalar/btree.rs"
],
"changed_files": 2,
"ci_check_suite_count": 6,
"c... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 1348,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-4235.test.patch",
"sha256": "082e32d3e218e7b50d1de125120199dcb8f2bb556fed2e0987ca25a5629330de",
"source_path": "data/lance-format__lance/benchmark_package... | {
"base_commit": "6e987921d0efbe42c018047ea45d1b6f624d8280",
"commands": [
{
"command": "cd python && /opt/venv/bin/python -m maturin develop --extras tests && /opt/venv/bin/python -m pytest -vvv -s -m 'not recurring and not integration and not slow' python/tests/test_scalar_index.py",
"command_inli... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "e6aaa4bd66cbec9bf3bd0333e4db4d46640faaa6e36649b83881610004f74707",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "c51aef5b2f... | 66e137982bc2dfb3dfe5e880f5e9efd372e78ab5 | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": false,
"has_tests_changed": false,
"language_area": "rust",
"language_areas": [
"rust"
],
"module_area": "query_scan",
"module_areas": [
"query_scan",
"dataset_table"
],
"patch_size_bucket": "small",
... | lance-format__lance-4309 | {
"combined_selected": true,
"combined_selection_rank": 8,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 8,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir": "... | 4,309 | https://github.com/lance-format/lance/pull/4309 | {
"excluded_sources": [
"linked_issue",
"tests"
],
"source_strategy": "pr_body",
"text": "When running SQL queries against a Lance dataset/table, an aggregate query of the form:\n\n- `SELECT count(*) FROM <table>`\n\ncurrently fails with an internal error instead of returning a single-row result.\n\nThi... | {
"bytes": 4251,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-4309.source.patch",
"private_artifact": true,
"sha256": "57025e002e3a6d1601a0081f78a42abfe8d49a396365d9dd53290ff653eb4e0d",
"source_path": "data/lance-format__lance/benc... | lance-format/lance | benchmark-instance-v1 | {
"additions": 82,
"author_association": "CONTRIBUTOR",
"base_commit": "26e00c67123275fbac6e53c6a3df29971b47ab92",
"base_ref": "main",
"changed_file_paths": [
"rust/lance-datafusion/src/projection.rs",
"rust/lance/src/datafusion/dataframe.rs",
"rust/lance/src/dataset/scanner.rs",
"rust/lance/s... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 2030,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-4309.test.patch",
"sha256": "64a91e124fa14a8db55d79c3ee13831d38f96969f3381eff4396bf9b4c4351bf",
"source_path": "data/lance-format__lance/benchmark_package... | {
"base_commit": "66e137982bc2dfb3dfe5e880f5e9efd372e78ab5",
"commands": [
{
"command": "cargo +stable test -p lance test_sql_count",
"command_inline_test_names": [
"test_sql_count"
],
"command_mode": "per_inline_test_name",
"language": "rust",
"matched_changed_files"... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "0278cbf0108c642e503ca0a21b092a34fc3ab730e40e6b4709e1ac4764901beb",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "485a7dab6a... | 0c691448a9e160cbd49979dc363347d1e07be4d8 | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": true,
"has_tests_changed": false,
"language_area": "rust",
"language_areas": [
"rust"
],
"module_area": "dataset_table",
"module_areas": [
"dataset_table",
"file_format_encoding"
],
"patch_size_bucket"... | lance-format__lance-4431 | {
"combined_selected": true,
"combined_selection_rank": 9,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 9,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir": "... | 4,431 | https://github.com/lance-format/lance/pull/4431 | {
"excluded_sources": [
"tests"
],
"source_strategy": "linked_issue",
"text": "Writing a Lance dataset using the V2.1 file format can panic for certain nested schemas when the data is highly compressible.\n\n## Current behavior\nWhen writing data with a nested structure like `struct<field: list<struct<...>>... | {
"bytes": 2230,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-4431.source.patch",
"private_artifact": true,
"sha256": "32f67cf6649b32b386ef3c3aec42f49b1259b106d24f2c825c7792c774c0a86d",
"source_path": "data/lance-format__lance/benc... | lance-format/lance | benchmark-instance-v1 | {
"additions": 157,
"author_association": "COLLABORATOR",
"base_commit": "0c691448a9e160cbd49979dc363347d1e07be4d8",
"base_ref": "main",
"changed_file_paths": [
"rust/lance-encoding/src/encodings/physical/rle.rs",
"rust/lance/src/dataset.rs"
],
"changed_files": 2,
"ci_check_suite_count": 13,
"... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 6610,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-4431.test.patch",
"sha256": "c9a626ece1d48ae73dbf303ae2c998994d1952540d9d535280156611c4dd2cf9",
"source_path": "data/lance-format__lance/benchmark_package... | {
"base_commit": "0c691448a9e160cbd49979dc363347d1e07be4d8",
"commands": [
{
"command": "cargo +stable test -p lance test_issue_4429_nested_struct_encoding_v2_1_with_over_65k_structs",
"command_inline_test_names": [
"test_issue_4429_nested_struct_encoding_v2_1_with_over_65k_structs"
],... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "2aefa8676c2ae0265688434d57201ef1ddb6298bfc92eff62319cd76aaba7383",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "f3530ca6c9... | 97a6a5d7a0fb20fbf8b8f3e7655b12b192a8cac7 | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": false,
"has_tests_changed": false,
"language_area": "rust",
"language_areas": [
"rust"
],
"module_area": "index_search",
"module_areas": [
"index_search"
],
"patch_size_bucket": "small",
"problem_source"... | lance-format__lance-4490 | {
"combined_selected": true,
"combined_selection_rank": 10,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 10,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir":... | 4,490 | https://github.com/lance-format/lance/pull/4490 | {
"excluded_sources": [
"labels",
"linked_issue",
"tests"
],
"source_strategy": "title_body",
"text": "After recent caching changes, full-text search / inverted index lookups can incorrectly reuse cached posting lists across different index partitions. When the same token ID is read from another par... | {
"bytes": 2106,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-4490.source.patch",
"private_artifact": true,
"sha256": "ccd2210fa3edee04eabb797e89ba19c667380b3e183b3819d4512116b02dab02",
"source_path": "data/lance-format__lance/benc... | lance-format/lance | benchmark-instance-v1 | {
"additions": 118,
"author_association": "CONTRIBUTOR",
"base_commit": "97a6a5d7a0fb20fbf8b8f3e7655b12b192a8cac7",
"base_ref": "main",
"changed_file_paths": [
"rust/lance-index/src/scalar/inverted/index.rs"
],
"changed_files": 1,
"ci_check_suite_count": 6,
"ci_conclusion": "success",
"ci_failed... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 5430,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-4490.test.patch",
"sha256": "ecde738a0563f506420d8b6ae83c0a20a8c0d1d0a7a54a519976e133298bb24f",
"source_path": "data/lance-format__lance/benchmark_package... | {
"base_commit": "97a6a5d7a0fb20fbf8b8f3e7655b12b192a8cac7",
"commands": [
{
"command": "cargo +stable test -p lance-index test_posting_cache_conflict_across_partitions",
"command_inline_test_names": [
"test_posting_cache_conflict_across_partitions"
],
"command_mode": "per_inline... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "8f34996a54c2adb22049a33246aa4a67e83e253e879da832d0a3f726dc211a3d",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "c0846edd30... | 76a710eb8e2efdbfe2cce9394606d360a020e7ec | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": true,
"has_tests_changed": true,
"language_area": "mixed",
"language_areas": [
"rust",
"python"
],
"module_area": "index_search",
"module_areas": [
"index_search",
"python_api",
"dataset_table"
]... | lance-format__lance-4667 | {
"combined_selected": true,
"combined_selection_rank": 11,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 11,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir":... | 4,667 | https://github.com/lance-format/lance/pull/4667 | {
"excluded_sources": [
"tests"
],
"source_strategy": "linked_issue",
"text": "Lance currently supports creating btree (scalar) indexes, but index creation can be impractically slow and memory-intensive on very large datasets when executed from a single process/node.\n\nAdd support for creating a btree inde... | {
"bytes": 57117,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-4667.source.patch",
"private_artifact": true,
"sha256": "e9aaa27a94c05670361be3fcc55a75f7990c5a842b3d720b81cc4cf425e65137",
"source_path": "data/lance-format__lance/ben... | lance-format/lance | benchmark-instance-v1 | {
"additions": 1569,
"author_association": "CONTRIBUTOR",
"base_commit": "76a710eb8e2efdbfe2cce9394606d360a020e7ec",
"base_ref": "main",
"changed_file_paths": [
"java/.gitignore",
"python/python/lance/dataset.py",
"python/python/lance/lance/__init__.pyi",
"python/python/tests/test_scalar_index... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 12968,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-4667.test.patch",
"sha256": "93323d70c21ef96cfbc8986ee3ba00772b5f2c61641c31a3eb1d711efb867f2f",
"source_path": "data/lance-format__lance/benchmark_packag... | {
"base_commit": "76a710eb8e2efdbfe2cce9394606d360a020e7ec",
"commands": [
{
"command": "cd python && /opt/venv/bin/python -m maturin develop --extras tests && /opt/venv/bin/python -m pytest -vvv -s -m 'not recurring and not integration and not slow' python/tests/test_scalar_index.py",
"command_inli... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "52fbd0402166ca491e0e3f0aa052b5d61c5cadaa1d67b60d61f7e87a915dc208",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "514761c055... | 419ac973396cdf09686c81532385e679b15b7e08 | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": false,
"has_tests_changed": true,
"language_area": "mixed",
"language_areas": [
"python",
"rust"
],
"module_area": "python_api",
"module_areas": [
"python_api",
"file_format_encoding"
],
"patch_siz... | lance-format__lance-4722 | {
"combined_selected": true,
"combined_selection_rank": 12,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 12,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir":... | 4,722 | https://github.com/lance-format/lance/pull/4722 | {
"excluded_sources": [
"linked_issue",
"tests"
],
"source_strategy": "title_body",
"text": "In the Python API, users may create a `LanceFileSession` configured with a particular object store (for example, with custom credentials, caching, or other options). However, there is currently no public way to ... | {
"bytes": 6870,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-4722.source.patch",
"private_artifact": true,
"sha256": "014ec32c8da10857904aac0a1ae6e924f8758fdef83d35da1c0d2c3eeb05c279",
"source_path": "data/lance-format__lance/benc... | lance-format/lance | benchmark-instance-v1 | {
"additions": 133,
"author_association": "CONTRIBUTOR",
"base_commit": "419ac973396cdf09686c81532385e679b15b7e08",
"base_ref": "main",
"changed_file_paths": [
"python/python/lance/file.py",
"python/python/lance/lance/__init__.pyi",
"python/python/tests/test_file.py",
"python/src/file.rs"
],... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 944,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-4722.test.patch",
"sha256": "ca56b2209dd98412ea6eeb9d21b1c9a10a66a23bc0bfe68a394926ac4646a916",
"source_path": "data/lance-format__lance/benchmark_package_... | {
"base_commit": "419ac973396cdf09686c81532385e679b15b7e08",
"commands": [
{
"command": "cd python && /opt/venv/bin/python -m maturin develop --extras tests && /opt/venv/bin/python -m pytest -vvv -s -m 'not recurring and not integration and not slow' python/tests/test_file.py",
"command_inline_test_... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "baa680750a3168eca16395fffe55106704036d322b4609b768672f2c37f9f080",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "178af79d62... | 94bb0400ac80ad8753b242eca93bb3f92509bfaa | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": false,
"has_tests_changed": true,
"language_area": "mixed",
"language_areas": [
"java",
"rust"
],
"module_area": "java_api",
"module_areas": [
"java_api"
],
"patch_size_bucket": "medium",
"problem_so... | lance-format__lance-4764 | {
"combined_selected": true,
"combined_selection_rank": 13,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 13,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir":... | 4,764 | https://github.com/lance-format/lance/pull/4764 | {
"excluded_sources": [
"labels",
"linked_issue",
"tests"
],
"source_strategy": "title_body",
"text": "In the Lance Java API, users need a way to supply a dataset manifest that they have already serialized (e.g., obtained from another process or cached locally) and then use that manifest to work wit... | {
"bytes": 6902,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-4764.source.patch",
"private_artifact": true,
"sha256": "2169bbfaab8b692732a9928f21937c5a584fd50f5ea517a15319e4622adcbd47",
"source_path": "data/lance-format__lance/benc... | lance-format/lance | benchmark-instance-v1 | {
"additions": 76,
"author_association": "CONTRIBUTOR",
"base_commit": "39eefea0044870ada03557bd2db116a85ad34877",
"base_ref": "main",
"changed_file_paths": [
"java/lance-jni/src/blocking_dataset.rs",
"java/src/main/java/com/lancedb/lance/Dataset.java",
"java/src/main/java/com/lancedb/lance/ReadOp... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 2910,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-4764.test.patch",
"sha256": "347836717585f5ef7ce1ff8b5507fc2364c97264241f90b24a8370e4dc25aa13",
"source_path": "data/lance-format__lance/benchmark_package... | {
"base_commit": "94bb0400ac80ad8753b242eca93bb3f92509bfaa",
"commands": [
{
"command": "cd java && ./mvnw test -Dtest=DatasetTest",
"command_inline_test_names": [],
"command_mode": "per_test_file",
"language": "java",
"matched_changed_files": [
"java/src/main/java/com/lanc... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "ce79f0eebbff85d9ba5982a65fd5026c35b797080763bf32f1fe02c13401f27d",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "f1bac6b201... | e3fa30c1d0422f972c0925f2e4fb136bd961139d | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": false,
"has_tests_changed": false,
"language_area": "rust",
"language_areas": [
"rust"
],
"module_area": "query_scan",
"module_areas": [
"query_scan",
"dataset_table"
],
"patch_size_bucket": "medium",
... | lance-format__lance-5019 | {
"combined_selected": true,
"combined_selection_rank": 14,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 14,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir":... | 5,019 | https://github.com/lance-format/lance/pull/5019 | {
"excluded_sources": [
"linked_issue",
"tests"
],
"source_strategy": "pr_body",
"text": "When running a scan/query over a Lance dataset with a SQL filter that combines `regexp_match(column, ...)` with additional boolean predicates using `AND`/`OR` (for example, checks like `col IS NOT NULL` and `col <>... | {
"bytes": 4883,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-5019.source.patch",
"private_artifact": true,
"sha256": "1ec40e5ac13db5d73d1dca71e4309b20c6aab4144cdaddbea6c08796ba54cb31",
"source_path": "data/lance-format__lance/benc... | lance-format/lance | benchmark-instance-v1 | {
"additions": 177,
"author_association": "CONTRIBUTOR",
"base_commit": "e3fa30c1d0422f972c0925f2e4fb136bd961139d",
"base_ref": "main",
"changed_file_paths": [
"rust/lance-datafusion/src/logical_expr.rs",
"rust/lance-datafusion/src/planner.rs",
"rust/lance/src/dataset/scanner.rs"
],
"changed_f... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 3560,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-5019.test.patch",
"sha256": "f70552bbdb59d41193d1e07a847805ac9d5684c9f8522564a08342b5252fb684",
"source_path": "data/lance-format__lance/benchmark_package... | {
"base_commit": "e3fa30c1d0422f972c0925f2e4fb136bd961139d",
"commands": [
{
"command": "cargo +stable test -p lance-datafusion test_regexp_match_and_non_empty_captions",
"command_inline_test_names": [
"test_regexp_match_and_non_empty_captions"
],
"command_mode": "per_inline_test... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "14fa41bca614df929c9422b3c82e77ec8fda2d0e86d66193a58352a254ec82c1",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "ced4673e3a... | 633aaa5422664f62f4e0a195948bdbee60e4d9db | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": true,
"has_tests_changed": true,
"language_area": "mixed",
"language_areas": [
"rust",
"python"
],
"module_area": "index_search",
"module_areas": [
"index_search",
"python_api",
"dataset_table",
... | lance-format__lance-5027 | {
"combined_selected": true,
"combined_selection_rank": 15,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 15,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir":... | 5,027 | https://github.com/lance-format/lance/pull/5027 | {
"excluded_sources": [
"tests"
],
"source_strategy": "linked_issue",
"text": "When using the Lance Python API with a dataset whose Arrow schema contains nested struct fields, creating an index on a nested field via dot notation does not work.\n\nReproduction example:\n- Create a PyArrow table with a schema... | {
"bytes": 21360,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-5027.source.patch",
"private_artifact": true,
"sha256": "08978e01632315c9e11221e9d6fd9e95072ad84b181816b2589a69ba1606bdad",
"source_path": "data/lance-format__lance/ben... | lance-format/lance | benchmark-instance-v1 | {
"additions": 739,
"author_association": "CONTRIBUTOR",
"base_commit": "add6fa69952c00568352fc12553113017ee128db",
"base_ref": "main",
"changed_file_paths": [
"python/python/lance/dataset.py",
"python/python/tests/test_scalar_index.py",
"python/python/tests/test_vector_index.py",
"python/src/... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 13834,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-5027.test.patch",
"sha256": "6694ed561ed6866b5cea85bd83d37b572afeef6991f15b3fa3728cc568bd815b",
"source_path": "data/lance-format__lance/benchmark_packag... | {
"base_commit": "633aaa5422664f62f4e0a195948bdbee60e4d9db",
"commands": [
{
"command": "cd python && /opt/venv/bin/python -m maturin develop --extras tests && /opt/venv/bin/python -m pytest -vvv -s -m 'not recurring and not integration and not slow' python/tests/test_scalar_index.py::test_nested_field_bt... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "feb99443798511e32fcec85006e3c9203e249666764f57a91c8380def22ae4b1",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "aca4414fa8... | 0662d11f69157b309acf024ac53d14d376dec578 | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": true,
"has_tests_changed": false,
"language_area": "rust",
"language_areas": [
"rust"
],
"module_area": "file_format_encoding",
"module_areas": [
"file_format_encoding"
],
"patch_size_bucket": "medium",
... | lance-format__lance-5074 | {
"combined_selected": true,
"combined_selection_rank": 16,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 16,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir":... | 5,074 | https://github.com/lance-format/lance/pull/5074 | {
"excluded_sources": [
"tests"
],
"source_strategy": "linked_issue",
"text": "Writing and then reading certain datasets using Lance’s storage format version V2.1 can crash at runtime.\n\nA concrete case is the Public BI Benchmark “CMSProvider” dataset. When this dataset is written with `WriteParams::with_s... | {
"bytes": 9964,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-5074.source.patch",
"private_artifact": true,
"sha256": "9f425854a7e55b9ff9c468a9a8c75b5833f437fdf210f3946d7506ee0112497a",
"source_path": "data/lance-format__lance/benc... | lance-format/lance | benchmark-instance-v1 | {
"additions": 129,
"author_association": "MEMBER",
"base_commit": "0662d11f69157b309acf024ac53d14d376dec578",
"base_ref": "main",
"changed_file_paths": [
"rust/lance-encoding/src/encodings/logical/primitive.rs",
"rust/lance-encoding/src/encodings/logical/primitive/blob.rs",
"rust/lance-encoding/s... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 2613,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-5074.test.patch",
"sha256": "27f82af751674e7a381aa736683dfd05596daf087ca4c1f9a0dd2d8bce69b6d6",
"source_path": "data/lance-format__lance/benchmark_package... | {
"base_commit": "0662d11f69157b309acf024ac53d14d376dec578",
"commands": [
{
"command": "cargo +stable test -p lance-encoding test_mixed_unraveler",
"command_inline_test_names": [
"test_mixed_unraveler"
],
"command_mode": "per_inline_test_name",
"language": "rust",
"m... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "159e005c880e3262810e751485973069cf8b22ac17556befc06dfbc08935f4bd",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "29fe449470... | f653be315a1883c95e88336a16bf542362263266 | {
"ci_conclusion": "failure",
"external_dependency_risk": false,
"has_linked_issue": false,
"has_tests_changed": true,
"language_area": "rust",
"language_areas": [
"rust"
],
"module_area": "io_object_store",
"module_areas": [
"io_object_store"
],
"patch_size_bucket": "medium",
"problem_s... | lance-format__lance-5077 | {
"combined_selected": true,
"combined_selection_rank": 17,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 17,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir":... | 5,077 | https://github.com/lance-format/lance/pull/5077 | {
"excluded_sources": [
"labels",
"linked_issue",
"tests"
],
"source_strategy": "pr_body",
"text": "When working with a Lance dataset that supports concurrent transactions, a conflict can be reported even when two operations are logically independent.\n\nSpecifically, if one transaction creates an i... | {
"bytes": 4889,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-5077.source.patch",
"private_artifact": true,
"sha256": "cb13bdfc1fe39a5c631a3f205f6d0cbd7cccbcf9412c1aa2902600e7d23f47f2",
"source_path": "data/lance-format__lance/benc... | lance-format/lance | benchmark-instance-v1 | {
"additions": 125,
"author_association": "MEMBER",
"base_commit": "f653be315a1883c95e88336a16bf542362263266",
"base_ref": "main",
"changed_file_paths": [
"python/python/tests/test_table_ops.py",
"rust/lance/src/io/commit/conflict_resolver.rs"
],
"changed_files": 2,
"ci_check_suite_count": 6,
... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 4056,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-5077.test.patch",
"sha256": "00ffa3c91b899825576a657f808dc1f0beb8d1332a7e7e55d6d80727e079e0b9",
"source_path": "data/lance-format__lance/benchmark_package... | {
"base_commit": "f653be315a1883c95e88336a16bf542362263266",
"commands": [
{
"command": "cd python && /opt/venv/bin/python -m maturin develop --extras tests && /opt/venv/bin/python -m pytest -vvv -s -m 'not recurring and not integration and not slow' python/tests/test_table_ops.py",
"command_inline_... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "047359b0f099a1cfbe660c67f4e3a98a393ee1b3c9bebafddce7b1bd4feb4836",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "bb32d471c9... | 772c6d87a5b1d9530b5f3772fba7c33cf5c55b3a | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": true,
"has_tests_changed": true,
"language_area": "rust",
"language_areas": [
"rust"
],
"module_area": "dataset_table",
"module_areas": [
"dataset_table",
"io_object_store"
],
"patch_size_bucket": "med... | lance-format__lance-5338 | {
"combined_selected": true,
"combined_selection_rank": 18,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 18,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir":... | 5,338 | https://github.com/lance-format/lance/pull/5338 | {
"excluded_sources": [
"pr_body",
"tests"
],
"source_strategy": "linked_issue",
"text": "When running the dataset cleanup procedure, it currently scans the entire dataset root and may treat unrelated files as candidates for deletion. This is problematic for datasets that intentionally store large exter... | {
"bytes": 12632,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-5338.source.patch",
"private_artifact": true,
"sha256": "900cdb2bb607cb916a0c1ef4433eb81c83a385eb3029714974471b7b5bc8717e",
"source_path": "data/lance-format__lance/ben... | lance-format/lance | benchmark-instance-v1 | {
"additions": 124,
"author_association": "CONTRIBUTOR",
"base_commit": "772c6d87a5b1d9530b5f3772fba7c33cf5c55b3a",
"base_ref": "main",
"changed_file_paths": [
"rust/lance-table/src/io/commit.rs",
"rust/lance-table/src/io/deletion.rs",
"rust/lance/src/dataset.rs",
"rust/lance/src/dataset/clean... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 1109,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-5338.test.patch",
"sha256": "29a6bb9f3b9a254e36e11665056cb446e8c7ef49038e2d8d68a1a01f156155d1",
"source_path": "data/lance-format__lance/benchmark_package... | {
"base_commit": "772c6d87a5b1d9530b5f3772fba7c33cf5c55b3a",
"commands": [
{
"command": "cargo +stable test -p lance cleanup_preserves_unmanaged_dirs_and_files",
"command_inline_test_names": [
"cleanup_preserves_unmanaged_dirs_and_files"
],
"command_mode": "per_inline_test_name",... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "0c11deb83b7b1d520ec3e066ef8537a1eaf4930c2f5265c96e12a4575cbb8f59",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "f8ba1651f7... | 8948169e6f9299721c0dd74aa01d565d225a4afa | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": true,
"has_tests_changed": false,
"language_area": "rust",
"language_areas": [
"rust"
],
"module_area": "dataset_table",
"module_areas": [
"dataset_table"
],
"patch_size_bucket": "small",
"problem_source... | lance-format__lance-5369 | {
"combined_selected": true,
"combined_selection_rank": 19,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 19,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir":... | 5,369 | https://github.com/lance-format/lance/pull/5369 | {
"excluded_sources": [
"tests"
],
"source_strategy": "linked_issue",
"text": "Lance Change Data Feed (CDF) incremental reading currently lets users specify a begin version and an end version to retrieve changes. However, users also need to retrieve incremental *upsert* changes (both newly inserted rows and... | {
"bytes": 4278,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-5369.source.patch",
"private_artifact": true,
"sha256": "4e7911a344a9dc7a3e1851846ee5b2d2cf3795281e8f0207a25d0ac4f6914b0b",
"source_path": "data/lance-format__lance/benc... | lance-format/lance | benchmark-instance-v1 | {
"additions": 135,
"author_association": "CONTRIBUTOR",
"base_commit": "8948169e6f9299721c0dd74aa01d565d225a4afa",
"base_ref": "main",
"changed_file_paths": [
"rust/lance/src/dataset/delta.rs"
],
"changed_files": 1,
"ci_check_suite_count": 8,
"ci_conclusion": "success",
"ci_failed_check_suite_c... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 2564,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-5369.test.patch",
"sha256": "53b56aff655cd3c577ff6f7c8f787ee8bd6abcecc974ef9346e494c333341f86",
"source_path": "data/lance-format__lance/benchmark_package... | {
"base_commit": "8948169e6f9299721c0dd74aa01d565d225a4afa",
"commands": [
{
"command": "cargo +stable test -p lance test_get_upsert_rows",
"command_inline_test_names": [
"test_get_upsert_rows"
],
"command_mode": "per_inline_test_name",
"language": "rust",
"matched_ch... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "0d0e0aac3ee7b6a87640b4d6f1167fe1a59e4a3c66fe4aec2815b114e04d1f56",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "45eea78e0f... | e7379a0fa5a397fea4cbccae5d0601fc682b2928 | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": false,
"has_tests_changed": false,
"language_area": "rust",
"language_areas": [
"rust"
],
"module_area": "file_format_encoding",
"module_areas": [
"file_format_encoding",
"dataset_table"
],
"patch_size... | lance-format__lance-5385 | {
"combined_selected": true,
"combined_selection_rank": 20,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 20,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir":... | 5,385 | https://github.com/lance-format/lance/pull/5385 | {
"excluded_sources": [
"linked_issues",
"tests"
],
"source_strategy": "pr_body",
"text": "Lance currently supports blobs stored directly within the dataset/file format, but users also need to store blob data externally (e.g., referenced by location/handle rather than embedded bytes) and still use Lance... | {
"bytes": 22805,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-5385.source.patch",
"private_artifact": true,
"sha256": "c4f70ba7520958d53b9896eed042e28d254c2fe17622d348e1324301e30b1075",
"source_path": "data/lance-format__lance/ben... | lance-format/lance | benchmark-instance-v1 | {
"additions": 278,
"author_association": "COLLABORATOR",
"base_commit": "e7379a0fa5a397fea4cbccae5d0601fc682b2928",
"base_ref": "main",
"changed_file_paths": [
"rust/lance-core/src/datatypes.rs",
"rust/lance-core/src/datatypes/field.rs",
"rust/lance-encoding/src/encodings/logical/blob.rs",
"r... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 2824,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-5385.test.patch",
"sha256": "07e3bbf300a66f98c33cbfa0c0e6ff4cb6395fce6b16bb81844ef27f5478a4db",
"source_path": "data/lance-format__lance/benchmark_package... | {
"base_commit": "e7379a0fa5a397fea4cbccae5d0601fc682b2928",
"commands": [
{
"command": "cargo +stable test -p lance-encoding test_blob_v2_external_round_trip",
"command_inline_test_names": [
"test_blob_v2_external_round_trip"
],
"command_mode": "per_inline_test_name",
"lan... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "091eeaa4690b7c0d9f83ac2f75f8112be2c53785589798c4444a8f928ba645fe",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "6e1022c233... | e7379a0fa5a397fea4cbccae5d0601fc682b2928 | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": false,
"has_tests_changed": false,
"language_area": "rust",
"language_areas": [
"rust"
],
"module_area": "dataset_table",
"module_areas": [
"dataset_table"
],
"patch_size_bucket": "small",
"problem_sourc... | lance-format__lance-5386 | {
"combined_selected": true,
"combined_selection_rank": 21,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 21,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir":... | 5,386 | https://github.com/lance-format/lance/pull/5386 | {
"excluded_sources": [
"linked_issues",
"tests"
],
"source_strategy": "title_body",
"text": "When updating a Lance dataset, the update process should not allow the blob version associated with existing data to change.\n\nCurrently, performing an update can incorrectly modify the blob version, even thou... | {
"bytes": 1953,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-5386.source.patch",
"private_artifact": true,
"sha256": "9130c7e88685625fa8b65fb181279e9837675e9dd4c0cdd376c7a982780b3090",
"source_path": "data/lance-format__lance/benc... | lance-format/lance | benchmark-instance-v1 | {
"additions": 32,
"author_association": "COLLABORATOR",
"base_commit": "e7379a0fa5a397fea4cbccae5d0601fc682b2928",
"base_ref": "main",
"changed_file_paths": [
"rust/lance/src/dataset/write.rs",
"rust/lance/src/dataset/write/insert.rs"
],
"changed_files": 2,
"ci_check_suite_count": 8,
"ci_conc... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 1365,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-5386.test.patch",
"sha256": "3365d41240dd188938ef23b301646d41fa824bc614db934ad2225cd6eb695c76",
"source_path": "data/lance-format__lance/benchmark_package... | {
"base_commit": "e7379a0fa5a397fea4cbccae5d0601fc682b2928",
"commands": [
{
"command": "cargo +stable test -p lance prevent_blob_version_upgrade_on_overwrite",
"command_inline_test_names": [
"prevent_blob_version_upgrade_on_overwrite"
],
"command_mode": "per_inline_test_name",
... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "ee109058eb55198807e781089ff8aa987ecdf48ce29be7c23f6ff0eb3ebd9e3f",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "e30d31b341... | eba080324a4594198f9dc224aa84fcc35c5c39ca | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": false,
"has_tests_changed": false,
"language_area": "rust",
"language_areas": [
"rust"
],
"module_area": "dataset_table",
"module_areas": [
"dataset_table",
"file_format_encoding"
],
"patch_size_bucket... | lance-format__lance-5406 | {
"combined_selected": true,
"combined_selection_rank": 22,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 22,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir":... | 5,406 | https://github.com/lance-format/lance/pull/5406 | {
"excluded_sources": [
"linked_issues (none)",
"tests (not provided)"
],
"source_strategy": "title_body",
"text": "Lance currently lacks dedicated support for storing and accessing blob (binary) values as a first-class feature in the dataset/file format APIs.\n\nAdd dedicated blob support so that users... | {
"bytes": 34673,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-5406.source.patch",
"private_artifact": true,
"sha256": "d69c3190d9f9006cc86d21584458f5cdb4ca08d1ec5bf7227091c97adb1ae949",
"source_path": "data/lance-format__lance/ben... | lance-format/lance | benchmark-instance-v1 | {
"additions": 504,
"author_association": "COLLABORATOR",
"base_commit": "eba080324a4594198f9dc224aa84fcc35c5c39ca",
"base_ref": "main",
"changed_file_paths": [
"rust/lance-arrow/src/schema.rs",
"rust/lance-core/src/datatypes/field.rs",
"rust/lance-core/src/utils.rs",
"rust/lance-core/src/util... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 897,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-5406.test.patch",
"sha256": "162dbcddbd04f31483a40de27edc1ef041085e6782ae65f948e1912817614544",
"source_path": "data/lance-format__lance/benchmark_package_... | {
"base_commit": "eba080324a4594198f9dc224aa84fcc35c5c39ca",
"commands": [
{
"command": "cargo +stable test -p lance-core blob_v2_detection_by_extension",
"command_inline_test_names": [
"blob_v2_detection_by_extension"
],
"command_mode": "per_inline_test_name",
"language": ... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "b5889e94970575ec904662a9daa1db4ae4c1be76db211616bcfcb05c971137b1",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "2e2087a901... | 6019650889268244c05f932e8dcc3adf6e522ff6 | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": false,
"has_tests_changed": false,
"language_area": "rust",
"language_areas": [
"rust"
],
"module_area": "versioning_transactions",
"module_areas": [
"versioning_transactions",
"namespace_catalog"
],
"... | lance-format__lance-5464 | {
"combined_selected": true,
"combined_selection_rank": 23,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 23,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir":... | 5,464 | https://github.com/lance-format/lance/pull/5464 | {
"excluded_sources": [
"linked_issue",
"tests"
],
"source_strategy": "pr_body",
"text": "When working with Lance datasets stored in an object store (cloud storage) using a directory-style namespace, constructing a table location by taking a root storage location and then appending a table/subpath can p... | {
"bytes": 4221,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-5464.source.patch",
"private_artifact": true,
"sha256": "1b1fa6d5aa0eadf4fce880472774d0de840ce4590fe8ab0f6c6d5080879a5486",
"source_path": "data/lance-format__lance/benc... | lance-format/lance | benchmark-instance-v1 | {
"additions": 74,
"author_association": "CONTRIBUTOR",
"base_commit": "6019650889268244c05f932e8dcc3adf6e522ff6",
"base_ref": "main",
"changed_file_paths": [
"rust/lance-namespace-impls/src/dir/manifest.rs"
],
"changed_files": 1,
"ci_check_suite_count": 6,
"ci_conclusion": "success",
"ci_failed... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 2656,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-5464.test.patch",
"sha256": "178a76f4410d49edaa0bc883dc448885c6da420bf5f1a5cdf692280bf9a8f58a",
"source_path": "data/lance-format__lance/benchmark_package... | {
"base_commit": "6019650889268244c05f932e8dcc3adf6e522ff6",
"commands": [
{
"command": "cargo +stable test -p lance-namespace-impls test_construct_full_uri_with_cloud_urls",
"command_inline_test_names": [
"test_construct_full_uri_with_cloud_urls"
],
"command_mode": "per_inline_t... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "f6d0764f29f3cff927aa4da047fc3fa1817dcf5c6149e3e8e141eef3fedb834e",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "a84687f5f2... | 303f58f161dd5dc4b6d8c40ebd08581f0aec1e6d | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": false,
"has_tests_changed": true,
"language_area": "mixed",
"language_areas": [
"python",
"rust"
],
"module_area": "python_api",
"module_areas": [
"python_api",
"dataset_table"
],
"patch_size_bucke... | lance-format__lance-5506 | {
"combined_selected": true,
"combined_selection_rank": 24,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 24,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir":... | 5,506 | https://github.com/lance-format/lance/pull/5506 | {
"excluded_sources": [
"labels",
"linked_issue",
"tests"
],
"source_strategy": "title_body",
"text": "Lance’s Python API supports scanning a dataset with filters and also supports both vector search and full-text search (FTS). However, it is currently not possible (or not exposed) to combine these ... | {
"bytes": 32625,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-5506.source.patch",
"private_artifact": true,
"sha256": "180ad32ed5a78a7bdbe41fb1808d88b4d9cc18d6c7a934fe16355ca9d5a56a7f",
"source_path": "data/lance-format__lance/ben... | lance-format/lance | benchmark-instance-v1 | {
"additions": 718,
"author_association": "CONTRIBUTOR",
"base_commit": "303f58f161dd5dc4b6d8c40ebd08581f0aec1e6d",
"base_ref": "main",
"changed_file_paths": [
"python/python/lance/dataset.py",
"python/python/lance/lance/__init__.pyi",
"python/python/tests/test_scalar_index.py",
"python/python... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 8680,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-5506.test.patch",
"sha256": "0dbadaf6a3c65cd5cda26c303c6b72ab66117156c271d26ea368f2212c9fe06d",
"source_path": "data/lance-format__lance/benchmark_package... | {
"base_commit": "303f58f161dd5dc4b6d8c40ebd08581f0aec1e6d",
"commands": [
{
"command": "cd python && /opt/venv/bin/python -m pip install --upgrade --force-reinstall 'lance-namespace>=0.4.5,<0.7' && /opt/venv/bin/python -m maturin develop --extras tests && /opt/venv/bin/python -m pip install --upgrade --f... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "91a04b434ebc616898ecd9673b90420216ffc47b20384d8c1f1dc8e3a753adb8",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "017db6722b... | d1ec03c5d95fc6abdd57393fb0eac2114eb54167 | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": false,
"has_tests_changed": true,
"language_area": "python",
"language_areas": [
"python"
],
"module_area": "dataset_table",
"module_areas": [
"dataset_table",
"python_api"
],
"patch_size_bucket": "sma... | lance-format__lance-5666 | {
"combined_selected": true,
"combined_selection_rank": 25,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 25,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir":... | 5,666 | https://github.com/lance-format/lance/pull/5666 | {
"excluded_sources": [
"linked_issue",
"tests"
],
"source_strategy": "pr_body",
"text": "In the Python API, configuring a scanner with `default_scan_options` should support passing vector search settings for `nearest` as a nested dictionary.\n\nCurrently, when `default_scan_options` is provided in a fo... | {
"bytes": 604,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-5666.source.patch",
"private_artifact": true,
"sha256": "c8e9b660aa9cba327000f3b58b71c5f5360e8676ca240ca8c4863d82eb5ae829",
"source_path": "data/lance-format__lance/bench... | lance-format/lance | benchmark-instance-v1 | {
"additions": 39,
"author_association": "CONTRIBUTOR",
"base_commit": "d1ec03c5d95fc6abdd57393fb0eac2114eb54167",
"base_ref": "main",
"changed_file_paths": [
"python/python/lance/dataset.py",
"python/python/tests/test_dataset.py"
],
"changed_files": 2,
"ci_check_suite_count": 4,
"ci_conclusio... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 1443,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-5666.test.patch",
"sha256": "8f4c64fb08edec3440312c88739813a4984c65f0fe65c304c38752e436a5de50",
"source_path": "data/lance-format__lance/benchmark_package... | {
"base_commit": "d1ec03c5d95fc6abdd57393fb0eac2114eb54167",
"commands": [
{
"command": "cd python && /opt/venv/bin/python -m pip install --upgrade --force-reinstall 'lance-namespace>=0.4.5,<0.7' && /opt/venv/bin/python -m maturin develop --extras tests && /opt/venv/bin/python -m pip install --upgrade --f... |
{
"combined_package": {
"generated_at": "2026-05-20T11:02:12",
"run_id": "lance_combined_50",
"source_instance_hash": "f8664e73361b1f2838f04c933d21bd711aba798bb02cf860e317513ce0545e27",
"source_package_dir": "data/lance-format__lance/benchmark_package_clean30"
},
"curated_record_hash": "211563766f... | 252424604130652fd1bbb43edc7142f2474c08ac | {
"ci_conclusion": "success",
"external_dependency_risk": false,
"has_linked_issue": false,
"has_tests_changed": true,
"language_area": "rust",
"language_areas": [
"rust"
],
"module_area": "dataset_table",
"module_areas": [
"dataset_table"
],
"patch_size_bucket": "small",
"problem_source... | lance-format__lance-6046 | {
"combined_selected": true,
"combined_selection_rank": 26,
"combined_source_set": "lance_format_clean30_all",
"packaged_at": "2026-05-20T11:02:12",
"selected": true,
"selection_rank": 26,
"selection_strategy": "lance_combined_50_all_lance_format_clean30_plus_lancedb_stratified20",
"source_package_dir":... | 6,046 | https://github.com/lance-format/lance/pull/6046 | {
"excluded_sources": [
"linked_issue",
"tests"
],
"source_strategy": "pr_body",
"text": "A filtered scan over a Lance dataset can crash (panic) in cases where the scan’s row selection mask is represented using bitmap-backed ranges and the mask spans multiple segments.\n\nThis can be triggered through n... | {
"bytes": 1160,
"contains_reference_solution": true,
"path": "data/lance_combined_50/private/reference_source_patches/lance-format__lance-6046.source.patch",
"private_artifact": true,
"sha256": "8058d130830e8f295a5b1348499387d9ca1a7ab5a46f1fd2d0a576bf8bcc8e0e",
"source_path": "data/lance-format__lance/benc... | lance-format/lance | benchmark-instance-v1 | {
"additions": 77,
"author_association": "CONTRIBUTOR",
"base_commit": "252424604130652fd1bbb43edc7142f2474c08ac",
"base_ref": "main",
"changed_file_paths": [
"rust/lance-table/src/rowids.rs",
"rust/lance/tests/query/primitives.rs"
],
"changed_files": 2,
"ci_check_suite_count": 7,
"ci_conclusi... | benchmark_packaging | {
"apply_before_evaluation": true,
"bytes": 3431,
"contains_reference_solution": false,
"path": "data/lance_combined_50/patches/tests/lance-format__lance-6046.test.patch",
"sha256": "be1d549e344b7d2e550d736f7b9a3082b46496004c3e506e23d15e6f1da35132",
"source_path": "data/lance-format__lance/benchmark_package... | {
"base_commit": "252424604130652fd1bbb43edc7142f2474c08ac",
"commands": [
{
"command": "cargo +stable test -p lance --features slow_tests --test integration_tests test_filtered_scan_after_compact_with_srid",
"command_inline_test_names": [],
"command_mode": "per_test_name",
"language": "... |
Lance Combined 50
This repository contains a 50-instance Lance/LanceDB software engineering benchmark and the final patch submissions from four agents. It is intended for reviewing task quality, verification evidence, and comparative model behavior on realistic Lance Format and LanceDB maintenance work.
TL;DR
- Lance Combined 50 is a verified 50-instance benchmark drawn from real Lance
and LanceDB issue/PR tasks: 30 from
lance-format/lanceand 20 fromlancedb/lancedb. - It is designed to test realistic maintenance work across APIs, query/index behavior, storage/dataset lifecycle, encoding/file format, and cross-language surfaces.
- Four agents were evaluated: GPT-5.5 Codex xhigh scored 22/50, Kimi K2.6 via OpenCode max reasoning scored 20/50, DeepSeek V4 Pro via OpenCode max reasoning scored 19/50, and Claude Code Opus 4.7 xhigh scored 18/50.
- All agents performed substantially better on
lancedb/lancedbthan onlance-format/lance, suggesting the Lance Format slice is harder and more lower-level/cross-layer. - The benchmark is not saturated: nine tasks are solved by all agents while 20 are missed by all; many misses fail at compile/type-check time, with the hard tail concentrated in lower-level Rust, encoding/blob/file-format, index semantics, Java API, and storage invariants.
What Is Included
| Path | Contents |
|---|---|
benchmark_instances.jsonl |
The 50 benchmark instances, one JSON object per task. |
benchmark_manifest.json |
Dataset provenance: selection counts, input hashes, selected ids, and packaging outputs. |
results/model_patches/claude_code_patches.json |
Claude Code Opus 4.7 xhigh final generated diffs for all 50 instances. |
results/model_patches/codex_patches.json |
GPT-5.5 Codex xhigh final generated diffs for all 50 instances. |
results/model_patches/deepseek_v4_pro_patches.json |
DeepSeek V4 Pro via OpenCode, max reasoning, final generated diffs for all 50 instances. |
results/model_patches/kimi_k2_6_patches.json |
Kimi K2.6 via OpenCode, max reasoning, final generated diffs for all 50 instances. |
The public repo contains benchmark metadata, provenance, and model submissions. It does not include reference solution patch files, full model trajectories, or full evaluation logs.
Agents Evaluated
| Agent | Patch file |
|---|---|
| GPT-5.5 Codex xhigh | results/model_patches/codex_patches.json |
| Kimi K2.6 via OpenCode, max reasoning | results/model_patches/kimi_k2_6_patches.json |
| DeepSeek V4 Pro via OpenCode, max reasoning | results/model_patches/deepseek_v4_pro_patches.json |
| Claude Code Opus 4.7 xhigh | results/model_patches/claude_code_patches.json |
Benchmark Construction
The benchmark contains 50 verified instances:
- 30 from
lance-format/lance - 20 from
lancedb/lancedb
Every instance has recorded packaging evidence that the source patch applies, the test patch applies, the pre-patch verification fails, and the post-patch verification passes. Verification was run on Modal, and the exact verification commands are stored in each instance row.
Agent Performance
| Agent | Passed | Score |
|---|---|---|
| GPT-5.5 Codex xhigh | 22 / 50 | 44% |
| Kimi K2.6 via OpenCode, max reasoning | 20 / 50 | 40% |
| DeepSeek V4 Pro via OpenCode, max reasoning | 19 / 50 | 38% |
| Claude Code Opus 4.7 xhigh | 18 / 50 | 36% |
Patch outcome breakdown:
| Agent | Passed | Compile/type failure | Test failure | Patch conflict |
|---|---|---|---|---|
| GPT-5.5 Codex xhigh | 22 | 16 | 6 | 6 |
| Kimi K2.6 via OpenCode, max reasoning | 20 | 18 | 7 | 5 |
| DeepSeek V4 Pro via OpenCode, max reasoning | 19 | 20 | 7 | 4 |
| Claude Code Opus 4.7 xhigh | 18 | 15 | 9 | 8 |
This breakdown is an evaluation-derived view of patch quality. Passed means
the generated patch applied, built, and satisfied the verification checks.
Compile/type failure usually indicates an incomplete or inconsistent edit.
Test failure means the patch ran but did not meet the target behavior.
Patch conflict means the generated patch did not compose cleanly with the
benchmark test patch, so it is best read as composition friction rather than a
pure semantic failure.
Patch style patterns from the final diffs:
| Agent | Avg files / patch | Multi-file patches | Patches touching 5+ files | Avg changed lines | Median changed lines |
|---|---|---|---|---|---|
| GPT-5.5 Codex xhigh | 3.3 | 35 / 50 | 10 / 50 | 160 | 107 |
| Kimi K2.6 via OpenCode, max reasoning | 2.4 | 28 / 50 | 4 / 50 | 97 | 62 |
| DeepSeek V4 Pro via OpenCode, max reasoning | 2.4 | 27 / 50 | 5 / 50 | 71 | 39 |
| Claude Code Opus 4.7 xhigh | 2.5 | 25 / 50 | 4 / 50 | 124 | 59 |
GPT-5.5 Codex xhigh produces the broadest patches: more files per instance, more multi-file edits, and the largest typical diff size. On this benchmark that broader style appears helpful for cross-module tasks such as index/search behavior and API-to-core integration. DeepSeek V4 Pro via OpenCode is the most compact by changed-line volume, which likely contributes to fewer patch conflicts but also leaves more incomplete compile/type failures. Kimi K2.6 via OpenCode is moderately scoped. Claude Code Opus 4.7 xhigh is less often multi-file than Codex, but has several large outlier patches and the highest patch-conflict count.
What The Results Show
The benchmark is not saturated. The best agent solves 22 of 50, and all four agents remain in a narrow band from 18 to 22 solved instances. That makes the set useful for comparing model behavior, but the main signal is not just the top-line ranking; it is the pattern of what each agent can and cannot repair.
GPT-5.5 Codex xhigh has the strongest patch-quality profile: the most passing patches, the fewest test failures, and a relatively balanced failure mix. Kimi K2.6 via OpenCode is close in score but has more compile/type failures, which suggests more incomplete integration. DeepSeek V4 Pro via OpenCode has the fewest patch conflicts and strong storage/internal-Rust performance, but also the highest compile/type failure count. Claude Code Opus 4.7 xhigh has the lowest score here and the most patch conflicts and test failures, although its compile/type failure count is not the worst.
All agents have much higher success rates on lancedb/lancedb than on
lance-format/lance: the LanceDB slice ranges from 50-60% solved, while the
Lance Format slice ranges from 27-33%. This suggests the Lance Format portion is
the harder slice, with more lower-level Rust, encoding, storage, and
cross-language integration work.
The four agents are complementary. GPT-5.5 Codex xhigh has the strongest overall score and does especially well on index/search/query behavior. DeepSeek V4 Pro via OpenCode is strongest on dataset lifecycle and storage-oriented work. Kimi K2.6 via OpenCode is strongest on language SDK and public API wrapper tasks. Encoding, Arrow, file-format, blob, and Java-side changes remain difficult for all four.
The result distribution also shows a meaningful hard tail: nine tasks are solved by all four agents, while 20 tasks are missed by all four. That hard subset is likely the most useful diagnostic material for understanding where current models struggle on Lance/LanceDB: deeper Rust internals, storage invariants, cross-language API consistency, and changes that require precise integration across multiple layers.
The easiest shared successes are the nine tasks solved by all four agents. These are mostly localized API, query, schema, and scan correctness fixes with direct verification targets. Examples include nested vector-search options in the Python API, hybrid-search prefilter behavior, schema preservation during table creation/overwrite, nested boolean filter planning, and filtered scans after delete/compaction.
The all-agent misses are more concentrated in lower-level or cross-layer work. They include Lance V2.1 encoding/null handling, blob read/write and blob version semantics, repetition-index caching, distributed scalar-index creation, nested field indexing, combined full-text/vector scan filtering, logical-vs-physical index API semantics, Java dataset APIs, manifest naming migration, cleanup safety, cloud namespace path construction, boolean bitmap indexes, remote version-check behavior, and TLS hostname verification. In short, the hardest cases tend to require preserving storage/index invariants across multiple subsystems rather than making a localized API fix.
Subsystem-level pass counts from the local analysis:
| Primary area | Size | Claude Opus 4.7 | GPT-5.5 Codex | DeepSeek V4 Pro | Kimi K2.6 |
|---|---|---|---|---|---|
| Index/search/query | 16 | 5 | 9 | 4 | 7 |
| Dataset lifecycle/storage | 12 | 6 | 4 | 7 | 6 |
| Encoding/Arrow/file-format/blob | 10 | 2 | 4 | 3 | 2 |
| Namespace/remote/security/connectivity | 6 | 2 | 2 | 3 | 1 |
| Language SDK/public API wrappers | 6 | 3 | 3 | 2 | 4 |
How To Read The Dataset
Each row in benchmark_instances.jsonl includes:
instance_id,repo,base_commit,pr_number, andpr_urlproblem_statement.titleandproblem_statement.text- source PR metadata such as changed files, labels, CI status, and linked issues
- verification commands, timeouts, and recorded pre/post command results
- test-patch metadata and reference-solution metadata
- distribution tags such as language area, module area, task type, patch size, and verification plausibility
The files in results/model_patches/ are arrays of 50 entries. Each entry has
instance_id, base_commit, prefix, and patch. These are final generated
diffs only; they do not contain trajectories, tool-call logs, setup logs, or
evaluation stdout/stderr.
- Downloads last month
- 30