cagtayfabry / pydeps2env Goto Github PK

currently, to move a package over to the pip install section, one has to provide the exactly matching package and version requirements string

e.g. if numpy>=1.20 is specified as a dependency, simply passing --pip="numpy" will have no effect (which is not very stable)

For the requirements.txt output, not all pip syntax is supported, e.g. pydeps2env@ git+https://github.com/CagtayFabry/pydeps2env.git cannot be installed from requirements.txt using pip

While using this action, I forgot to prepend the checkout action. So the input file was missing and I got this exception:

    env["dependencies"] = ["python" + cp.get("options", "python_requires")]
  File "/usr/lib/python3.8/configparser.py", line 781, in get
    d = self._unify_values(section, vars)
  File "/usr/lib/python3.8/configparser.py", line 1149, in _unify_values
    raise NoSectionError(section) from None
configparser.NoSectionError: No section: 'options'

I suggest it be more useful to treat this error. E.g. the configparser creation should fail, if the input file is not present.

Some configuration options would be nice, on the top of my head:

• list of dependencies to install via pip
• mapping of dependency names that differ for pip and conda (e.g. weldx-widgets -> weldx_widgets)

add configuration to exclude certain requirements (including wildcards)

For consistency (and reliable caching with mamba-org/provision-with-micromamba or similar) the dependencies listed in the output environment file should be sorted

currently the dependency order can be random (I think this is based on the toml implementation when reading the config file)