The pip-run project defines a py-module that has a dash in the name. It does so intentionally in order to make it possible to type python -m pip-run without having to enter any chorded characters (- is roughly 1/3 as expensive as _).
After converting the config to pyproject.toml, however, this value is rejected by policy.
ValueError: invalid pyproject.toml config: `tool.setuptools.py-modules[0]`.
configuration error: `tool.setuptools.py-modules[0]` must be python-module-name
Backend: run command get_requires_for_build_editable with args {'config_settings': None}
configuration error: `tool.setuptools.py-modules[0]` must be python-module-name
GIVEN VALUE:
"pip-run"
OFFENDING RULE: 'format'
DEFINITION:
{
"type": "string",
"format": "python-module-name"
}
I can understand why "python-module-name" seemed like a good reasonableness check, but it turns out to be backward incompatible and unnecessarily constraining against a use-case like above.
Can we back out this change and make it lenient as it was with setup.cfg?
The pip-run project defines a py-module that has a dash in the name. It does so intentionally in order to make it possible to type
python -m pip-runwithout having to enter any chorded characters (-is roughly 1/3 as expensive as_).After converting the config to pyproject.toml, however, this value is rejected by policy.
I can understand why "python-module-name" seemed like a good reasonableness check, but it turns out to be backward incompatible and unnecessarily constraining against a use-case like above.
Can we back out this change and make it lenient as it was with setup.cfg?