Cleanup the legacy engines code path

[madsbk]

Test helpers (cudf_polars/testing/asserts.py)

• assert_collect_raises

Tests

• test_collect_assert_raises
• assert_collect_raises imports (in test_asserts.py,
  test_stringfunction.py, test_casting.py,
  test_datetime_basic.py)

Benchmark CLI (benchmarks/utils.py)

• Removed the --executor flag and folded its cpu and
  in-memory choices into --frontend, which now accepts:
  cpu, in-memory, spmd, ray, dask, and duckdb.
• ExecutorType alias
• RunConfig.executor
• Executor serialization / summary output
• assert_never(run_config.executor) branch
• Unused assert_never import

Benchmark shim

• benchmarks/utils_new_frontends.py
  (contents merged into benchmarks/utils.py)

[coderabbitai]

Note

Reviews paused

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

Walkthrough

This PR eliminates global default executor configuration from the cudf_polars testing infrastructure by making engine specification explicit throughout. It requires all assertion functions to accept explicit GPUEngine instances, removes the --executor CLI option and helper functions that provided defaults, updates the entire test suite to pass explicit engines, consolidates benchmark utilities from a separate module into the main utils module, and adds comprehensive CLI and execution infrastructure for benchmark runners.

Changes

Testing and Benchmark Engine Parameterization

Layer / File(s)	Summary
Assertion function API refactoring python/cudf_polars/cudf_polars/testing/asserts.py	assert_gpu_result_equal and assert_sink_result_equal now require an explicit engine: GPUEngine keyword-only parameter. Removed DEFAULT_EXECUTOR, get_default_engine(), and assert_collect_raises() to eliminate optional engine fallbacks.
Pytest configuration and test fixture cleanup python/cudf_polars/tests/conftest.py	Removed the pytest_addoption() hook that registered --executor CLI flag and the wiring in pytest_configure() that set DEFAULT_EXECUTOR from it. Configuration now only handles marker registration and global ResourceWarning suppression.
Test suite updates for explicit engines python/cudf_polars/tests/expressions/test_casting.py, python/cudf_polars/tests/expressions/test_datetime_basic.py, python/cudf_polars/tests/expressions/test_stringfunction.py, python/cudf_polars/tests/test_config.py, python/cudf_polars/tests/test_executors.py, python/cudf_polars/tests/test_sink.py, python/cudf_polars/tests/test_tracing.py, python/cudf_polars/tests/testing/test_asserts.py	All test modules replaced assert_collect_raises() calls and executor="..." string arguments with explicit pytest.raises(...) blocks or instantiated pl.GPUEngine(executor="in-memory", raise_on_fail=True) objects. Updated subprocess environment to use explicit CUDF_POLARS__EXECUTOR setting instead of relying on test defaults.
Benchmark utilities consolidation and CLI infrastructure python/cudf_polars/cudf_polars/experimental/benchmarks/utils.py (previously utils_new_frontends.py removed)	Merged benchmarking datamodels (QueryResult, RunConfig, validation records), hardware/version tracking (HardwareInfo, PackageVersions), frontend execution runners (CPU, in-memory, SPMD, Ray, Dask, DuckDB), NVTX-annotated timing, structured tracing collection, and comprehensive CLI argument parsing (build_parser, parse_args, run_polars) into a single, self-contained module. Includes dataset loading, scale-factor inference, validation golden-file discovery, and result summarization.

Estimated code review effort

🎯 4 (Complex) | ⏱️ ~60 minutes

The review involves understanding the systematic removal of a global executor pattern across multiple test files and assertion utilities, verifying that all test branches correctly specify engines, and validating the large consolidation of benchmark utilities that introduces comprehensive CLI infrastructure and multi-frontend execution orchestration.

Possibly related PRs

• rapidsai/cudf#22410: Related refactoring of default engine/executor handling in asserts.py with similar removal of fallback engine derivation patterns.
• rapidsai/cudf#22381: Extends test fixtures in conftest.py with Ray/Dask-aware streaming engine markers and fixtures that work in tandem with the explicit engine requirement pattern introduced here.

Suggested reviewers

• TomAugspurger
• mroeschke
• bdice
• Matt711

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 59.09% which is insufficient. The required threshold is 80.00%.	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 'Cleanup the legacy engines code path' accurately describes the main objective of removing deprecated engine-related code and making engine arguments mandatory in assertion functions.
Linked Issues check	✅ Passed	The PR successfully implements all coding requirements from issue #22346: makes engine argument mandatory for assert_gpu_result_equal and assert_sink_result_equal, removes DEFAULT_EXECUTOR and get_default_engine, eliminates assert_collect_raises, and removes the --executor CLI option to prevent implicit defaults.
Out of Scope Changes check	✅ Passed	All changes are directly aligned with the linked issue #22346 objectives. The refactoring of benchmarks/utils.py consolidating frontend selection and removal of utils_new_frontends.py are supporting changes to achieve explicit engine handling throughout the codebase.
Description check	✅ Passed	The PR description clearly outlines the removal of legacy engine code paths, test helpers, and CLI flag consolidation, matching the changeset which removes deprecated test utilities and refactors benchmark/testing infrastructure.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

Warning

Review ran into problems

🔥 Problems

Stopped waiting for pipeline failures after 30000ms. One of your pipelines takes longer than our 30000ms fetch window to run, so review may not consider pipeline-failure results for inline comments if any failures occurred after the fetch window. Increase the timeout if you want to wait longer or run a @coderabbit review after the pipeline has finished.

---

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

[madsbk]

This diff is hard to read on GitHub. To better isolate the changes in
utils.py, use something like:

git diff \
  main:python/cudf_polars/cudf_polars/experimental/benchmarks/utils_new_frontends.py \
  HEAD:python/cudf_polars/cudf_polars/experimental/benchmarks/utils.py

[wence-]

• --executor CLI option

  • pytest_addoption

  • DEFAULT_EXECUTOR = config.getoption("--executor")

This stuff has already gone, no?

[wence-]

Closes #22346.

This is closed by refusing to do it, AFAICT, rather than shoveling an engine argument into assert_ir_translation_raises.

[wence-]

Please update the PR description for what is now actually going on.

TBH, I would prefer just changing the benchmark rather than the benchmark and the tests in this PR

[wence-]

Why?

[madsbk]

Inlining it instead of calling assert_collect_raises is simpler IMO.

[madsbk]

Closes #22346.

This is closed by refusing to do it, AFAICT, rather than shoveling an engine argument into assert_ir_translation_raises.

Removed the closing, you are right that this PR only addresses the first part of #22346.
I am not sure assert_ir_translation_raises is really an issue, though.

[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.

[madsbk]

TBH, I would prefer just changing the benchmark rather than the benchmark and the tests in this PR

Fair point, let's do one at a time: #22504

[madsbk]

close in favor of #22535