Add ArrayData::new_null and DataType::primitive_width #3676
Conversation
tustvold
force-pushed
the
add-new-null-array
branch
from
836062e
to
be8bb83
Compare
tustvold
force-pushed
the
add-new-null-array
branch
from
be8bb83
to
9630674
Compare
tustvold
force-pushed
the
add-new-null-array
branch
from
febfba7
to
c156207
Compare
Co-authored-by: askoa <112126368+askoa@users.noreply.github.com>
| true, | ||
| ), | ||
| DataType::Union(f, i, mode) => { | ||
| let ids = Buffer::from_iter(std::iter::repeat(i[0]).take(len)); |
There was a problem hiding this comment.
Is it correct? Why repeating same type id? I think this indicates all values in the UnionArray are from same child array (id[0])? But looks like you still produce all children null array?
There was a problem hiding this comment.
I was being lazy 😆 It should be harmless for the children to have more data than they technically need.
There was a problem hiding this comment.
Ha. 😄
Just wondering if it will cause issue later, e.g. interoperate with other arrow libraries?
|
Benchmark runs are scheduled for baseline = 6ec7226 and contender = 7cd29d7. 7cd29d7 is a master commit associated with this PR. Results will be available as each benchmark for each run completes. |
Which issue does this PR close?
Closes #.
Rationale for this change
This makes it easier to create a null array of a concrete type, adds support for more types of null array (e.g. UnionArray and RunArray), and reduces some code duplication
What changes are included in this PR?
Are there any user-facing changes?