Skip to content

Commit de213c5

Browse files
singhpratechclaude
andcommitted
Upstream tracker: binary_slice default-stop overflow filed as apache/arrow#51222
Co-Authored-By: Claude Fable 5.1 <[email protected]>
1 parent 6d5d061 commit de213c5

1 file changed

Lines changed: 1 addition & 1 deletion

File tree

‎docs/UPSTREAM.md‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -12,7 +12,7 @@ the workaround can be removed.
1212
| pyarrow 25.0.1 | `pc.pairwise_diff` on a sliced array ignores `ArrowArray.offset` | `test_pyarrow_pairwise_diff_ignores_the_array_offset` | reported by others: [apache/arrow#50524](https://github.com/apache/arrow/issues/50524), fixed by [#50858](https://github.com/apache/arrow/pull/50858) | fixed in 26.0.0; the test retires when the venv moves to 26 |
1313
| pyarrow 25.0.1 | `pc.fill_null_forward`, `pc.fill_null_backward`, `pc.replace_with_mask` on a sliced boolean array ignore the offset | `test_pyarrow_boolean_fill_null_forward_ignores_the_array_offset` | not yet filed; distinct from [#45086](https://github.com/apache/arrow/issues/45086) (chunked-output sizing, fixed in 26.0.0) | open, reproduced 2026-09-07, draft ready |
1414
| pyarrow 25.0.1 | `pc.winsorize` on a sliced array nulls the wrong rows and places values where the input has nulls | `test_pyarrow_winsorize_ignores_the_array_offset` | not yet filed; no existing report found | open, reproduced 2026-09-07, draft ready |
15-
| pyarrow 25.0.1 | `pc.binary_slice` with its own default `stop` raises `ArrowInvalid: Negative buffer resize` | `test_pyarrow_binary_slice_overflows_on_its_own_default_stop` | not yet filed; the same overflow was fixed for `utf8_slice_codeunits` in [#36575](https://github.com/apache/arrow/pull/36575), the binary path was not; umbrella [#34929](https://github.com/apache/arrow/issues/34929) | open, reproduced 2026-09-07, draft ready |
15+
| pyarrow 25.0.1 | `pc.binary_slice` with its own default `stop` raises `ArrowInvalid: Negative buffer resize` | `test_pyarrow_binary_slice_overflows_on_its_own_default_stop` | filed 2026-09-08: [apache/arrow#51222](https://github.com/apache/arrow/issues/51222); the same overflow was fixed for `utf8_slice_codeunits` in [#36575](https://github.com/apache/arrow/pull/36575), the binary path was not; umbrella [#34929](https://github.com/apache/arrow/issues/34929) | filed, open |
1616
| pyarrow 25.0.1 | `pc.year_month_day` and `pc.iso_calendar` crashed the process in the first matrix run | the matrix never calls them; `temporal_struct` compares field by field ([EVALUATION.md](EVALUATION.md)) | withheld: not reproduced on 2026-09-07 under guard malloc | unexplained, observed once |
1717
| Arrow (vendored date library) | DST rules stop being applied after the last tabulated transition (2037), so a summer instant in 2050 gets the winter offset, on a fat TZif file | finding 14 in [EVALUATION.md](EVALUATION.md), 144 matrix cases | not yet filed; related [#42157](https://github.com/apache/arrow/issues/42157) (open) covers slim files only | open, reproduced 2026-09-07, draft ready |
1818
| Apple Metal | the Feature Set Tables say the Apple9 family has "the full set of 64-bit atomic operations", but the shading-language headers admit `device ulong` only for atomic min and max: no 64-bit add, compare-exchange, load or store compile, whatever language version is asked for (measured on an M4 Max, macOS 26.6.2). Grouped sums and hash tables carry 32-bit workarounds because of it | [DECISIONS.md](DECISIONS.md), [DESIGN.md](DESIGN.md); the probe is attached to the draft report | Feedback Assistant, not yet filed, draft ready | open |

0 commit comments

Comments
 (0)