Fix deprecation warning: declaring default stream as cuda::stream_ref

[davidwendt]

Description

Fixes a deprecation warning where cudf::stream_ref was variable was declared without a stream.

[285+64+832=1180] Building CXX object benchmarks/CMakeFiles/MERGE_NVBENCH.dir/merge/merge_lists.cpp.o
/cudf/cpp/benchmarks/merge/merge_lists.cpp: In function 'void nvbench_merge_list(nvbench::state&)':
/cudf/cpp/benchmarks/merge/merge_lists.cpp:17:20: warning: 'constexpr cuda::__4::stream_ref::stream_ref()' is deprecated: Using the default/null stream is generally discouraged. If you need to use it, please construct a stream_ref from cudaStream_t{nullptr} [-Wdeprecated-declarations]
   17 |   cuda::stream_ref stream;
      |                    ^~~~~~
In file included from /cudf/cpp/build/_deps/cccl-src/lib/cmake/libcudacxx/../../../libcudacxx/include/cuda/__stream/get_stream.h:26,
                 from /cudf/cpp/build/_deps/cccl-src/lib/cmake/libcudacxx/../../../libcudacxx/include/cuda/stream_ref:56,
                 from /conda/envs/rapids/include/rmm/cuda_stream_view.hpp:10,
                 from /cudf/cpp/include/cudf/utilities/default_stream.hpp:10,
                 from /cudf/cpp/include/cudf/column/column_view.hpp:8,
                 from /cudf/cpp/include/cudf/column/column.hpp:7,
                 from /cudf/cpp/include/cudf/table/table.hpp:7,
                 from /cudf/cpp/benchmarks/common/generate_nested_types.hpp:8,
                 from /cudf/cpp/benchmarks/merge/merge_lists.cpp:6:
/cudf/cpp/build/_deps/cccl-src/lib/cmake/libcudacxx/../../../libcudacxx/include/cuda/__stream/stream_ref.h:63:3: note: declared here
   63 |   stream_ref() = default;
      |   ^~~~~~~~~~
[280+64+837=1180] Building CXX object benchmarks/CMakeFiles/MERGE_NVBENCH.dir/merge/merge_structs.cpp.o
/cudf/cpp/benchmarks/merge/merge_structs.cpp: In function 'void nvbench_merge_struct(nvbench::state&)':
/cudf/cpp/benchmarks/merge/merge_structs.cpp:17:20: warning: 'constexpr cuda::__4::stream_ref::stream_ref()' is deprecated: Using the default/null stream is generally discouraged. If you need to use it, please construct a stream_ref from cudaStream_t{nullptr} [-Wdeprecated-declarations]
   17 |   cuda::stream_ref stream;
      |                    ^~~~~~
In file included from /cudf/cpp/build/_deps/cccl-src/lib/cmake/libcudacxx/../../../libcudacxx/include/cuda/__stream/get_stream.h:26,
                 from /cudf/cpp/build/_deps/cccl-src/lib/cmake/libcudacxx/../../../libcudacxx/include/cuda/stream_ref:56,
                 from /conda/envs/rapids/include/rmm/cuda_stream_view.hpp:10,
                 from /cudf/cpp/include/cudf/utilities/default_stream.hpp:10,
                 from /cudf/cpp/include/cudf/column/column_view.hpp:8,
                 from /cudf/cpp/include/cudf/column/column.hpp:7,
                 from /cudf/cpp/include/cudf/table/table.hpp:7,
                 from /cudf/cpp/benchmarks/common/generate_nested_types.hpp:8,
                 from /cudf/cpp/benchmarks/merge/merge_structs.cpp:6:
/cudf/cpp/build/_deps/cccl-src/lib/cmake/libcudacxx/../../../libcudacxx/include/cuda/__stream/stream_ref.h:63:3: note: declared here
   63 |   stream_ref() = default;
      |   ^~~~~~~~~~

Introduced by #23769

Checklist

• I am familiar with the Contributing Guidelines.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 88555e7f-9258-4238-aa35-981f50a99c16

📥 Commits

Reviewing files that changed from the base of the PR and between f54eb2a and 86fc1ba.

📒 Files selected for processing (2)

• cpp/benchmarks/merge/merge_lists.cpp
• cpp/benchmarks/merge/merge_structs.cpp

Included review availability: Your plan provides up to 12 included reviews per hour; 11 remain after this review.

---

📝 Walkthrough

Summary by CodeRabbit

• Bug Fixes
  • Improved stream handling and synchronization in merge benchmarks for more reliable benchmark execution.

Walkthrough

The merge list and merge struct benchmarks now explicitly use cuda::stream_ref for the default stream and call stream.sync().

Changes

Merge benchmark stream updates

Layer / File(s)	Summary
Default stream and synchronization updates cpp/benchmarks/merge/merge_lists.cpp, cpp/benchmarks/merge/merge_structs.cpp	Both benchmarks declare the default stream as cuda::stream_ref and use stream.sync() for synchronization.

Estimated code review effort: 1 (Trivial) | ~2 minutes

Merge Risk: ⚪ Minimal · up to 86fc1

This localized benchmark-only change removes a deprecation warning without introducing an actionable merge-blocking risk; it is merge-ready after normal checks and review.

Suggested reviewers: qbacpey, vyasr

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 2 functions across 2 files.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly identifies the main change: fixing a deprecation warning caused by the default construction of cuda::stream_ref in the merge benchmarks.
Description check	✅ Passed	The description directly explains the deprecation warning, identifies the affected benchmark files, and describes the stream declaration fix.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

• Fix all pre-merge checks with AI

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands.}

[bdice]

We need to fix this so cudf::get_default_stream() returns a cuda::stream_ref. Otherwise we'll be going backwards in our efforts to move away from rmm::cuda_stream_view here.

[davidwendt]

That seems like an ongoing change that would eventually fix this code. Can we merge this until then perhaps?

[davidwendt]

I've made a change here that should be valid even after the cudf::get_default_stream() is updated.

[mhaseeb123]

Looks good aside from discussion with @bdice, I am fine either way :)