[BUG] Usage of copy constructor in column.hpp breaks compilation

[wmalpica]

Describe the bug
Building blazingsql requires to include cudf/table/table.hpp which in turn includes cudf/column/column.hpp

After the changes in rmm https://github.com/rapidsai/rmm/pull/775/files
The copy constructor now requires providing a cuda_stream_view, which is not being provided in line 104:

include/cudf/column/column.hpp:104:36: error: use of deleted function ‘rmm::device_buffer::device_buffer(const rmm::device_buffer&)’
       _children{std::move(children)}

Steps/Code to reproduce bug
I am trying to update the blazingsql code to account for the changes in
https://github.com/rapidsai/rmm/pull/775/files
If you try to build my branch, you will see the same error:

git clone https://github.com/wmalpica/blazingsql.git
cd blazingsql
git checkout fix/update_rmm_device_buffer
./dependencies.sh 21.08 11.2 nightly

Expected behavior
It should build

Environment overview (please complete the following information)
Baremetal. cudf installed using conda. Versions in use:
cmake 3.18.5 h1f3970d_0 rapidsai-nightly
cudf 21.08.00a210616 cuda_11.2_py38_g91c727f396_166 rapidsai-nightly
dask-cuda 21.08.00a210615 py38_27 rapidsai-nightly
dask-cudf 21.08.00a210616 py38_g91c727f396_166 rapidsai-nightly
libcudf 21.08.00a210616 cuda11.2_g15b18f46e8_167 rapidsai-nightly
librmm 21.08.00a210615 cuda11.2_g91491aa_25 rapidsai-nightly
rmm 21.08.00a210615 cuda_11.2_py38_g91491aa_25 rapidsai-nightly
ucx 1.9.0+gcd9efd3 cuda11.2_0 rapidsai-nightly
ucx-proc 1.0.0 gpu rapidsai-nightly
ucx-py 0.20.0a210615 py38_gcd9efd3_41 rapidsai-nightly

Environment details
Please run and paste the output of the cudf/print_env.sh script here, to gather any other relevant environment details

Additional context
Add any other context about the problem here.

[harrism]

Hi @wmalpica, can you paste the line of code that calls the column ctor (line 104)? I suspect that you just need to wrap the argument to the null_mask parameter of that ctor with std::move. (note the std::forward in the implementation -- if you don't move it, the forward will result in a copy, which I suspect is what is calling the deleted device_buffer copy ctor.

I had to change a number of these in libcudf after changing rmm::device_buffer. For example:

https://github.com/rapidsai/cudf/pull/8280/files#diff-1140f9642e1a1dcc42c6830acb3f3681eb71f63f36cd7444cfb692e6c487777bL55-L57

Basically, rapidsai/rmm#775 uncovered a number of places where we were copying but should have been moving.

[wmalpica]

Thanks @harrism , you are correct, I was simply not moving the data. I was so focused on the changes to the cuda_stream_view not being set by default, I did not really see what was happening. Sorry to waste your time. This is great, now we too have noticed when we were inadvertently copying data.

[harrism]

No worries, happy to help!