[CUB] Refactor DeviceAdjacentDifference::SubtractLeft to always take an environment

[miscco]

Description

closes

Checklist

• New or existing tests cover these changes.
• The documentation is up to date with these changes.

[copy-pr-bot]

Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

[coderabbitai]

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

DeviceAdjacentDifference SubtractLeft and SubtractLeftCopy overloads switch stream parameters to const EnvT& env, route execution through dispatch_with_env_and_tuning with a policy selector, update convenience-enable_if to require indirectly-binary-invocable, and add iterator concepts plus Catch2 tests covering multiple environment types and temp-storage usage.

Changes

Environment-based dispatch for SubtractLeft APIs*

Layer / File(s)	Summary
Iterator binary invocability concepts libcudacxx/include/cuda/std/__iterator/concepts.h	Add indirectly_binary_invocable and indirectly_regular_binary_invocable concepts extending unary patterns to two iterators with copy-constructibility, invocability across value/reference/common-reference operand combinations, and common-reference result constraints; provide both _CCCL_HAS_CONCEPTS() concept definitions and fallback _CCCL_CONCEPT/_CCCL_REQUIRES_EXPR macro implementations.
Temp-storage variant API contract and dispatch cub/cub/device/device_adjacent_difference.cuh	Temp-storage SubtractLeftCopy and SubtractLeft replace cudaStream_t stream with const EnvT& env, add EnvT template parameter and API documentation, include cuda/std/__iterator/concepts.h, and route execution through dispatch_with_env_and_tuning into adjacent_difference::dispatch with the computed policy selector and left/read option.
Convenience overload signature and lambda forwarding cub/cub/device/device_adjacent_difference.cuh	Convenience overloads change SFINAE constraints to require ::cuda::std::indirectly_binary_invocable<...>, take const EnvT& env by const reference, and adjust dispatch_with_env_and_tuning lambdas to forward storage/bytes with the policy_selector into adjacent_difference::dispatch.
CUB integration tests for environment variants cub/test/catch2_test_device_adjacent_difference_substract_left.cu	Add CUDA standard execution headers and two Catch2 tests for SubtractLeft and SubtractLeftCopy that query temp-storage size with null calls, allocate user temp storage, run each API across multiple environment/policy variants (raw stream, cuda::stream, cuda::stream_ref, cuda::std::execution::env, cuda::execution::gpu), perform CUDA error checks and synchronization, and validate GPU outputs against std::adjacent_difference.
Iterator concept tests and support types libcudacxx/test/libcudacxx/std/iterators/..., libcudacxx/test/support/indirectly_readable.h	Add compile-pass tests for indirectly_binary_invocable and indirectly_regular_binary_invocable with positive cases (GoodInvocable overload sets covering value/reference/common-reference combinations) and negative cases (non-copy-constructible invocables, deleted operator() for specific argument categories, return-type common-reference failures); add IndirectlyReadable2 support type mirroring existing pattern for two-iterator tests.

Possibly related PRs

• NVIDIA/cccl#9318: Migrates other Device APIs to const EnvT& environment-based dispatch via dispatch_with_env_and_tuning.

Suggested reviewers

• gevtushenko
• shwina
• gonidelis

---

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

[coderabbitai]

Actionable comments posted: 2

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 0203dc71-7802-4f71-a286-b6a226094643

📥 Commits

Reviewing files that changed from the base of the PR and between c345dcf and ba420ef.

📒 Files selected for processing (2)

• cub/cub/device/device_adjacent_difference.cuh
• cub/test/catch2_test_device_adjacent_difference_substract_left.cu

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[github-actions]

🥳 CI Workflow Results

🟩 Finished in 2h 55m: Pass: 100%/287 | Total: 2d 20h | Max: 1h 15m | Hits: 99%/198283

See results here.