hcrosse opened a new pull request, #1587:
URL: https://github.com/apache/arrow-go/pull/1587
### Rationale for this change
`scalar.GetScalar` on a sliced `List`, `LargeList`, `Map` or `FixedSizeList`
counted rows from the buffer's start, ignoring the slice offset.
I reproduced this on v18.8.0 and main, with Go 1.26.10 on macOS arm64.
```go
full, _, _ := array.FromJSON(mem, arrow.FixedSizeListOf(2,
arrow.PrimitiveTypes.Int32),
strings.NewReader(`[[1, 2], [3, 4], [5, 6]]`))
sliced := array.NewSlice(full, 1, 3)
sc, _ := scalar.GetScalar(sliced, 0)
```
Both versions return `[1 2]` instead of `[3 4]`. A `list<list<int32>>` row
holds a slice of the child list array, so equal nested scalars can also hash
differently even when the caller slices nothing.
### What changes are included in this PR?
The four `GetScalar` cases now use `ValueOffsets(idx)`, which adds the array
offset, for both bounds, as Arrow C++ does through `value_slice`:
https://github.com/apache/arrow/blob/976779b2aac587f51b87c3bd2e6ae4c36f75ab6c/cpp/src/arrow/array/array_base.cc#L101-L106
This corrects nested `scalar.Append` and `scalar.Hash`, and extraction
through struct, sparse union, run-end encoded and dictionary values over sliced
list arrays.
### Are these changes tested?
`TestGetScalarSlicedListLike`, `TestListScalarHashNestedSliced` and
`TestAppendNestedSlicedListScalar` cover extraction, hashing and appending. All
three fail on main and pass with this change.
Full `go test -race ./arrow/...` and `golangci-lint` 2.13.2 passed on macOS
arm64 with pinned test data.
### Are there any user-facing changes?
Yes. `scalar.GetScalar` returns the correct row from sliced list arrays.
There is no API change.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]