PP301 check fails with pytest 9's native TOML configuration

[tsvikas]

Issue

The PP301 check requires [tool.pytest.ini_options] in pyproject.toml, but pytest 9 introduced native TOML support under [tool.pytest] ([PR #13743](pytest-dev/pytest#13743)).

Using the new format causes false failure:

├── PP301 Has pytest in pyproject ❌
│   Must have a [tool.pytest.ini_options] configuration section in pyproject.toml.
├── PP302 Sets a minimum pytest to at least 6 [skipped]
├── PP303 Sets the test paths [skipped]
...

Expected

Checks should accept either:

• [tool.pytest.ini_options] (INI-compatibility mode, pytest 6+)
• [tool.pytest] (native TOML, pytest 9+)

Note: Both tables cannot be used simultaneously per pytest documentation.

References

• [pytest 9.0 changelog](https://docs.pytest.org/en/stable/changelog.html#pytest-9-0-0-2024-12-21)
• [pytest configuration docs](https://docs.pytest.org/en/stable/reference/customize.html#pyproject-toml)

[tsvikas]

also note that in pytest 9, xfail_strict, strict-config, and strict-markers can all be enabled with strict = true (relevant for PP305-307)

[TimothyEDawson]

@henryiii Not sure if this deserves its own issue, but I noticed my CI just started failing on the validate-pyproject step, and I couldn't reproduce it on my local system so I had a tough time figuring out exactly what caused it.

Eventually I figured out that it's because the legacy schema for [tool.pytest.ini_options] (https://www.schemastore.org/partial-pytest.json) includes the following:

"xfail_strict": {
  "type": "string",
  "enum": ["true", "false"],
  "description": "DEPRECATED. Alias for strict_xfail. If true, test suite will fail if a test marked with `@pytest.mark.xfail` unexpectedly passes.",
  "deprecated": true,
  "default": "false"
}

Meanwhile the copier template produces a pyproject.toml with xfail_strict = true. I changed this to xfail_strict = "true" and this seems to have resolved the problem.

[henryiii]

That's incorrect, bools are also supported here. That probably is a bug when the schema was reworked.

[henryiii]

(The schema is incorrect, that is :) )

[TimothyEDawson]

I figured as much, given I've had no issues using a bool for quite a while now! I don't know who maintains that schema or when exactly it was broken, but it should probably be reported to them.

[henryiii]

I'm hoping someone contributes the fix, but I could probably do it eventually. It's pretty easy to edit.

[henryiii]

And this is breaking our CI too, since we check the schema.

[henryiii]

I've yanked validate-pyproject-schema-store 2025.11.13, hopefully the fix will be in the next release.