[libcu++] Backport `cuda::std::constant_wrapper` to C++17 by davebayer · Pull Request #10687 · NVIDIA/cccl

davebayer · 2026-08-06T13:09:19Z

No description provided.

coderabbitai · 2026-08-06T13:15:27Z

Note

Reviews paused

It looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the reviews.auto_review.auto_pause_after_reviewed_commits setting.

Use the following commands to manage reviews:

@coderabbitai resume to resume automatic reviews.
@coderabbitai review to trigger a single review.

Use the checkboxes below for quick actions:

▶️ Resume reviews
🔍 Trigger review

📝 Walkthrough

Summary by CodeRabbit

New Features
- Enabled constant-wrapper functionality across configured language standards, including C++17, while retaining NVRTC limitations.
- Expanded support for arithmetic, logical, comparison, conversion, invocation, subscript, and kernel-parameter use cases.
Bug Fixes
- Improved constant-wrapper operations and type detection in pre-C++20 environments.
Tests
- Expanded compatibility coverage across language standards.
- Added validation for operator behavior, conversions, overloads, runtime interactions, and compiler-specific cases.

Walkthrough

Changes

The constant_wrapper implementation now supports configured language standards except NVRTC. Its tests support C++17 where possible, replace C++20 concepts with SFINAE traits, and guard C++20-only cases.

Constant-wrapper compatibility

Layer / File(s)	Summary
Implementation guard and unary address-of `libcudacxx/include/cuda/std/__utility/constant_wrapper.h`	The implementation is enabled outside NVRTC. Unary `operator&` uses C++17-compatible syntax.
Baseline wrapper tests `libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/{adl,assign,comma,convert,cw,general,kernel_parameter,kernel_parameter.array,mem_ptr,types.compile}.pass.cpp`	Tests no longer exclude C++17 when supported. C++20-only checks remain conditionally compiled.
Operator tests `libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/{unary_ops,binary_ops,comp,pseudo_mutators}.pass.cpp`	Operator concepts were replaced with SFINAE-based detection traits. Result type checks use separate `decltype` and `same_as` assertions.
Invocation and subscript tests `libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/{call,subscript}.pass.cpp`	Invocation and subscript coverage separates C++17-compatible cases from C++20-only cases and uses explicit template syntax where required.

Possibly related PRs

NVIDIA/cccl#9702: Both changes modify constant_wrapper.h and its associated tests.
NVIDIA/cccl#10588: Both changes modify constant_wrapper.h and its associated tests.

Suggested reviewers: fbusato, miscco

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

coderabbitai

Actionable comments posted: 2

🧹 Nitpick comments (2)

libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/unary_ops.pass.cpp (1)

121-185: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win

suggestion: Rename the detection templates to the required naming convention. Rename T to a PascalCase name such as Type. Rename HasPlus, HasMinus, HasBitNot, HasNot, HasBitAnd, HasDeref, and the HasNoexcept* templates to snake_case names.

As per coding guidelines, use PascalCase for template parameter names and snake_case for all other symbols.

Source: Coding guidelines
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/binary_ops.pass.cpp (1)
192-346: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚖️ Poor tradeoff

suggestion: collapse the 24 detection templates with a macro.

Each pair differs only by the operator token. One macro removes about 150 lines of copy-paste and keeps the SFINAE key and the noexcept expression in sync by construction.
+#define DEFINE_BINARY_OP_TRAITS(Name, Op)                                                                     \
+  template <class L, class R, class = void>                                                                    \
+  inline constexpr bool Has##Name = false;                                                                     \
+  template <class L, class R>                                                                                  \
+  inline constexpr bool                                                                                        \
+    Has##Name<L, R, cuda::std::void_t<decltype(cuda::std::declval<L&>() Op cuda::std::declval<R&>())>> = true;  \
+  template <class L, class R, class = void>                                                                    \
+  inline constexpr bool HasNoexcept##Name = false;                                                              \
+  template <class L, class R>                                                                                  \
+  inline constexpr bool                                                                                        \
+    HasNoexcept##Name<L, R, cuda::std::void_t<decltype(cuda::std::declval<L&>() Op cuda::std::declval<R&>())>> = \
+      noexcept(cuda::std::declval<L&>() Op cuda::std::declval<R&>())
+
+DEFINE_BINARY_OP_TRAITS(Plus, +);
+DEFINE_BINARY_OP_TRAITS(Minus, -);
+// ... remaining operators

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 8e16e497-0482-4403-8cc3-71de68ee7573

📥 Commits

Reviewing files that changed from the base of the PR and between fef5a52 and 3f25828.

📒 Files selected for processing (17)

libcudacxx/include/cuda/std/__utility/constant_wrapper.h
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/adl.compile.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/assign.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/binary_ops.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/call.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/comma.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/comp.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/convert.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/cw.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/general.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/kernel_parameter.array.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/kernel_parameter.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/mem_ptr.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/pseudo_mutators.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/subscript.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/types.compile.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/unary_ops.pass.cpp

💤 Files with no reviewable changes (2)

libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/kernel_parameter.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/kernel_parameter.array.pass.cpp

coderabbitai · 2026-08-06T13:25:39Z

Note

GitHub couldn't provide a complete incremental comparison for this pull request, so CodeRabbit is performing a full review instead. This review may take a little longer.

coderabbitai

Actionable comments posted: 1

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 7d3e900f-fcd1-4f15-a0ad-5fff3e3fd565

📥 Commits

Reviewing files that changed from the base of the PR and between fef5a52 and 5aa6ac6.

📒 Files selected for processing (17)

libcudacxx/include/cuda/std/__utility/constant_wrapper.h
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/adl.compile.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/assign.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/binary_ops.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/call.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/comma.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/comp.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/convert.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/cw.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/general.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/kernel_parameter.array.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/kernel_parameter.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/mem_ptr.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/pseudo_mutators.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/subscript.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/types.compile.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/unary_ops.pass.cpp

💤 Files with no reviewable changes (2)

libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/kernel_parameter.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/kernel_parameter.array.pass.cpp

🚧 Files skipped from review as they are similar to previous changes (13)

libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/adl.compile.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/mem_ptr.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/comma.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/convert.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/binary_ops.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/types.compile.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/unary_ops.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/cw.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/call.pass.cpp
libcudacxx/include/cuda/std/__utility/constant_wrapper.h
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/comp.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/general.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/subscript.pass.cpp

coderabbitai · 2026-08-06T13:30:31Z

Note

GitHub couldn't provide a complete incremental comparison for this pull request, so CodeRabbit is performing a full review instead. This review may take a little longer.

coderabbitai · 2026-08-06T13:32:25Z

Note

GitHub couldn't provide a complete incremental comparison for this pull request, so CodeRabbit is performing a full review instead. This review may take a little longer.

coderabbitai

Actionable comments posted: 2

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 777000c5-7fb0-486e-bb49-80e7b851eb0a

📥 Commits

Reviewing files that changed from the base of the PR and between fef5a52 and 216af9c.

📒 Files selected for processing (17)

libcudacxx/include/cuda/std/__utility/constant_wrapper.h
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/adl.compile.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/assign.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/binary_ops.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/call.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/comma.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/comp.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/convert.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/cw.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/general.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/kernel_parameter.array.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/kernel_parameter.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/mem_ptr.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/pseudo_mutators.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/subscript.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/types.compile.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/unary_ops.pass.cpp

💤 Files with no reviewable changes (2)

libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/kernel_parameter.array.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/kernel_parameter.pass.cpp

🚧 Files skipped from review as they are similar to previous changes (11)

libcudacxx/include/cuda/std/__utility/constant_wrapper.h
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/types.compile.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/mem_ptr.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/adl.compile.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/general.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/unary_ops.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/binary_ops.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/call.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/comma.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/convert.pass.cpp
libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/subscript.pass.cpp

github-actions · 2026-08-07T08:00:37Z

⏱️ CCCL compile-time benchmark comparison: Public headers compile-time bench

Result: 2 regression row(s), 12 improvement row(s) above threshold.

Run	Value
Config	`public-headers-gcc13`
Baseline	`origin/main`
Preset	`all-dev`
Targets	`cub.headers.base, thrust.cpp.cuda.headers.base, libcudacxx.test.public_headers`
GPU / launch args	`rtx2080` / `--cuda 13.3 --host gcc13`

Artifacts: reports and traces

TU total compilation

-f total-compilation inclusive --sort total

🔴 TU total compilation — Regressions

Rank	Regression impact	Selected Δ	Baseline	Current	Event	Matched traces
1	`4.639144`	`4.639144`	`13.415309`	`18.054453`	Total Compilation Time: `thrust/shuffle.h`	1
2	`3.155232`	`3.155232`	`14.370129`	`17.525361`	Total Compilation Time: `thrust/transform_reduce.h`	1

🟢 TU total compilation — Improvements

Rank	Improvement impact	Selected Δ	Baseline	Current	Event	Matched traces
1	`4.359860`	`-4.359860`	`20.227235`	`15.867375`	Total Compilation Time: `cuda/execution`	1
2	`3.738350`	`-3.738350`	`10.380363`	`6.642013`	Total Compilation Time: `cub/agent/agent_merge_sort.cuh`	1
3	`3.727775`	`-3.727775`	`20.328296`	`16.600521`	Total Compilation Time: `cuda/std/execution`	1
4	`3.252869`	`-3.252869`	`14.525175`	`11.272306`	Total Compilation Time: `cub/device/device_copy.cuh`	1

Direct file processing

-f file-processing exclusive --sort total

🟢 Direct file processing — Improvements

Rank	Improvement impact	Selected Δ	Baseline	Current	Event	Matched traces
1	`1.583696`	`-1.583696`	`23.816642`	`22.232946`	Processing Header File: `libcudacxx/include/cuda/__device/physical_device.h`	94
2	`1.468507`	`-1.468507`	`10.432210`	`8.963703`	Processing Header File: `libcudacxx/include/cuda/std/__cccl/prologue.h`	550
3	`0.479162`	`-0.479162`	`3.690283`	`3.211121`	Processing Header File: `libcudacxx/include/cuda/std/__cccl/epilogue.h`	550
4	`0.466628`	`-0.466628`	`6.291381`	`5.824753`	Processing Header File: `libcudacxx/include/cuda/std/__iterator/iterator_traits.h`	464
5	`0.450448`	`-0.450448`	`12.747601`	`12.297153`	Processing Header File: `libcudacxx/include/cuda/std/__type_traits/type_list.h`	477
6	`0.427840`	`-0.427840`	`12.115479`	`11.687639`	Processing Header File: `libcudacxx/include/cuda/std/__iterator/concepts.h`	437
7	`0.226635`	`-0.226635`	`1.888419`	`1.661784`	Processing Header File: `libcudacxx/include/cuda/std/__limits/numeric_limits_ext.h`	409
8	`0.202125`	`-0.202125`	`2.336037`	`2.133912`	Processing Header File: `cub/cub/device/dispatch/tuning/tuning_scan.cuh`	68

github-actions · 2026-08-07T09:39:56Z

🥳 CI Workflow Results

🟩 Finished in 2h 12m: Pass: 100%/115 | Total: 1d 01h | Max: 1h 11m | Hits: 99%/327444

See results here.

davebayer requested a review from a team as a code owner August 6, 2026 13:09

davebayer requested a review from fbusato August 6, 2026 13:09

github-project-automation Bot added this to CCCL Aug 6, 2026

github-project-automation Bot moved this to Todo in CCCL Aug 6, 2026

cccl-authenticator-app Bot moved this from Todo to In Review in CCCL Aug 6, 2026

coderabbitai Bot reviewed Aug 6, 2026

View reviewed changes

Comment thread libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/assign.pass.cpp Outdated

Comment thread libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/cw.pass.cpp Outdated

davebayer force-pushed the constant_wrapper_cpp17 branch from 3f25828 to 5aa6ac6 Compare August 6, 2026 13:25

This comment has been minimized.

Sign in to view

davebayer force-pushed the constant_wrapper_cpp17 branch from 5aa6ac6 to 31a7623 Compare August 6, 2026 13:27

coderabbitai Bot reviewed Aug 6, 2026

View reviewed changes

Comment thread libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/pseudo_mutators.pass.cpp Outdated

[libcu++] Backport cuda::std::constant_wrapper to C++17

216af9c

davebayer force-pushed the constant_wrapper_cpp17 branch from 31a7623 to 216af9c Compare August 6, 2026 13:32

coderabbitai Bot reviewed Aug 6, 2026

View reviewed changes

Comment thread libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/comp.pass.cpp

Comment thread libcudacxx/test/libcudacxx/std/utilities/const.wrap.class/comp.pass.cpp

This comment has been minimized.

Sign in to view

davebayer added 2 commits August 6, 2026 16:55

fix

b956aa0

fix old msvc

3d51414

This comment has been minimized.

Sign in to view

davebayer added 2 commits August 7, 2026 08:20

fix

93cf380

and another one

ad81429

This comment has been minimized.

Sign in to view

fix

4f9096e

This comment has been minimized.

Sign in to view

miscco approved these changes Aug 7, 2026

View reviewed changes

davebayer enabled auto-merge (squash) August 7, 2026 08:44

davebayer merged commit 5200422 into NVIDIA:main Aug 7, 2026
142 checks passed

github-project-automation Bot moved this from In Review to Done in CCCL Aug 7, 2026

This was referenced Aug 7, 2026

[libcu++] Use static call operator when possible #10711

Open

[libcu++] Fix nvcc picking up hidden friends on windows #10723

Open

Uh oh!

Conversation

davebayer commented Aug 6, 2026

Uh oh!

coderabbitai Bot commented Aug 6, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Reviews paused

Summary by CodeRabbit

Walkthrough

Changes

Uh oh!

coderabbitai Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

coderabbitai Bot commented Aug 6, 2026

Uh oh!

This comment has been minimized.

coderabbitai Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

coderabbitai Bot commented Aug 6, 2026

Uh oh!

coderabbitai Bot commented Aug 6, 2026

Uh oh!

coderabbitai Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

github-actions Bot commented Aug 7, 2026

⏱️ CCCL compile-time benchmark comparison: Public headers compile-time bench

TU total compilation

Direct file processing

Uh oh!

github-actions Bot commented Aug 7, 2026

🥳 CI Workflow Results

🟩 Finished in 2h 12m: Pass: 100%/115 | Total: 1d 01h | Max: 1h 11m | Hits: 99%/327444

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

coderabbitai Bot commented Aug 6, 2026 •

edited

Loading