Reading csvy model causes problems using Configuration

[wkerzendorf]

Code Sample, a copy-pastable example if possible

config = Configuration.from_yaml('blondin_model_compare_ddc10.yml')
run_tardis(config)

Problem description

This produces an error about v_inner_boundary not allowed to be defined. Just using the file directly works.

Added by Jordi:

Using a similar procedure with a different file produces a different error:

config = Configuration.from_yaml('tardis_csvy_example.yml')
run_tardis(config)

jsonschema.exceptions.ValidationError: Additional properties are not allowed ('config_dirname' was unexpected)

If the dictionary is being generated by TARDIS itself this shouldn't be a problem.
I'm looking into both problems

[ebjordi]

I'm looking into it

[ebjordi]

when you run run_tardis(config) TARDIS tries to create a Configuration object from that config arg without verifying if it's already one, it considers that is a filename or a simple dict:

     36         tardis_config = Configuration.from_yaml(config)
     37     except TypeError:
---> 38         tardis_config = Configuration.from_config_dict(config)

and checking Configuration.from_yaml it end up using Configuration.from_config_dict

    def from_yaml(cls, fname, *args, **kwargs):
        try:
            yaml_dict = yaml_load_file(
                fname, loader=kwargs.pop("loader", YAMLLoader)
            )
        except IOError as e:
            logger.critical(f"No config file named: {fname}")
            raise e

        tardis_config_version = yaml_dict.get("tardis_config_version", None)
        if tardis_config_version != "v1.0":
            raise ConfigurationError(
                "Currently only tardis_config_version v1.0 supported"
            )

        kwargs["config_dirname"] = os.path.dirname(fname)

        return cls.from_config_dict(yaml_dict, *args, **kwargs)

so the implementation becomes redundant, maybe the solution would be to add a verification before trying to make a new Configuration object?