jhorstmann commented on pull request #9329:
URL: https://github.com/apache/arrow/pull/9329#issuecomment-768208269
How cheap cloning of the ArrayData struct is probably depends a lot on the
datatype. The removed indirection seems to be beneficial, but instead we now
need to clone all fields, which include
- the datatype enum, which can include boxes/vecs of nested types and the
field metadata
- the vec of buffers, and cloning a buffer would include cloning the
`Arc<Bytes>`
So for primitive types, removing the arc should be a benefit because of less
indirection when accessing it, it would still need to allocates vectors and
clone the Arc which is inside the Buffer.
For more complex types, cloning the DataType enum itself and multiple
buffers could be slower than the speedup gained by less indirection.
----------------------------------------------------------------
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]
