jhorstmann edited a comment on pull request #9211:
URL: https://github.com/apache/arrow/pull/9211#issuecomment-761764712
@jorgecarleitao @alamb The testcase I had is not yet fixed, for reference
it's [ARROW-11267] with
```
#[test]
fn test_list_different_offsets() {
let a = create_list_array(&[Some(&[0, 0]), Some(&[1, 2]), Some(&[3,
4])]);
let b = create_list_array(&[Some(&[1, 2]), Some(&[3, 4]), Some(&[5,
6])]);
let a_slice = a.slice(1, 2);
let b_slice = b.slice(0, 2);
test_equal(&a_slice, &b_slice, true);
}
```
I'm currently thinking it's related to the `child_logical_null_buffer` and
`logical_list_bitmap` functions. One problem with that implementation is that
it creates an all-valid bitmap as a default when there is no null bitmap in
order to simplify the following calculations. This newly created bitmap then
has an offset of 0, while for actual nullable arrays the offset is that of the
parent data parameter. Maybe one approach would be to always pass tuples of
`(offset, Buffer)` for the null buffers.
[1]: https://issues.apache.org/jira/browse/ARROW-11267
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
