Closed hzuo closed 1 year ago
Patch coverage has no change and project coverage change: -0.03
:warning:
Comparison is base (
38a2e90
) 83.66% compared to head (702eaba
) 83.63%.
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Do you have feedback about the report comment? Let us know in this issue.
LGTM.
BTW, The Buffer
is copy free, so it's free to call array.values().clone()
Yeap, seems good. :+1:
Many kernels operate directly on the underlying buffer, creating a new allocation for a new buffer with updated values, and re-use the other components of the array like the offsets, bitmap, etc.
In many cases, the the output of the kernel is transient, and the allocation for the buffer can be reused - as we run the kernel over a big set of arrays, we can reuse the same scratch buffer (similar to the scratch buffers used in the io modules).
Exposing
into_inner
allows applications to re-use the scratch allocation instead of calling out to the allocator for every array.PrimitiveArray
s andBooleanArray
s already have aninto_inner
which deconstruct themselves into their constituent components - here we just bring Utf8Arrays and BinaryArrays in line with the other array types.