[pylaterreur]

Description

Follow-up to #12313.

#12313 capped pytest below 9.2 in the cuda.cccl.examples lanes: pytest 9.2.0 removed the private Function._nodeid, and pytest-benchmark 5.3.0 read it in its benchmark fixture (ionelmc/pytest-benchmark#321), so every host-benchmark case errored during setup. pytest-benchmark 5.3.1 (2026-11-01, ionelmc/pytest-benchmark#323) reads the public nodeid instead.

As the TODO in #12313 asks, this PR removes the pytest cap from the Linux and Windows install lines and requires pytest-benchmark>=5.3.1.

Checked locally on Python 3.14 (no GPU), with a parametrized test that uses the benchmark fixture under --benchmark-disable, as the smoke test does:

pytest	pytest-benchmark	Result
9.2.0	5.3.0	2 errors, AttributeError: 'Function' object has no attribute '_nodeid', as in the CI logs
9.2.0	5.3.1	2 passed
9.1.1	5.3.0	2 passed (the current cap)

• pip install --dry-run pytest pytest-xdist "pytest-benchmark>=5.3.1" resolves to pytest 9.2.0, pytest-xdist 3.8.0 and pytest-benchmark 5.3.1.
• Current CI already installs pytest-benchmark 5.3.1, together with the capped pytest 9.1.1.
• pre-commit run --files passes on both scripts.

Not verified: the cuda.cccl.examples CI lanes themselves.

Checklist

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

🤖 Generated with Claude Code

https://claude.ai/code/session_01GwUvhmzgfkiZ6vo4uGyesN

[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.

[coderabbitai]

No actionable comments were generated in the recent review.

Summary

Summary by CodeRabbit

• Chores
  • Updated the examples test setup on Linux and Windows to support newer pytest versions.
  • Raised the minimum supported version of the benchmark testing tool; pytest is no longer restricted to versions below 9.2.

Both Linux and Windows examples test setup scripts remove the pytest version cap and require pytest-benchmark version 5.3.1 or newer.

Changes: Examples test dependencies

Layer / File(s)	Summary
Update test install requirements ci/util/python/run_examples_tests.sh, ci/windows/run_examples_tests.ps1	Both scripts remove the pytest<9.2 constraint and add pytest-benchmark>=5.3.1. The Windows script also removes the related TODO.

Suggested reviewers: gonidelis

Change: Other

Merge Risk: Minimal · up to 653a3

This updates CI test dependency ranges so newer pytest versions work with pytest-benchmark. It does not affect production behavior, and merge risk is minimal. The examples CI lanes should confirm it after merge.

Architecture risk: Low · up to 653a3

The change affects 1 system.

Changed systems: ci

Architecture concerns
No architecture-level concerns identified.

Review details

Systems and components

• observed — ci (service) was modified; 2 changed files map to changed impact.

Before / after behavior

• observed — Modified behavior in ci/util/python/run_examples_tests.sh: The install command removes the pytest<9.2 constraint and replaces the unpinned pytest-benchmark requirement with pytest-benchmark>=5.3.1.
• observed — Modified behavior in ci/windows/run_examples_tests.ps1: The install step removes the pytest<9.2 cap and the corresponding TODO, and adds a pytest-benchmark minimum version of 5.3.1 while retaining upgrades of pip, pytest, and pytest-xdist.

- [ ] Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Recent review info

Run configuration

• Source mode: Auto
• Repository: NVIDIA/cccl
• Branch: main
• File: .coderabbit.yaml
• Commit: 5d63af4
• Layers: Repository YAML
• Review profile: CHILL
• Plan: Enterprise
• Run ID: b88f4601-8cef-441f-ba89-8a951380669e

Commits

Reviewing files that changed from the base of the PR and between 5d63af4 and 653a3c5.

Files selected for processing: 2

• ci/util/python/run_examples_tests.sh
• ci/windows/run_examples_tests.ps1

Included review availability: This review used your included allowance. Your plan provides up to 12 included reviews per hour; 10 remain after this review.