`environment-name` now is a mandatory argument

[ocefpaf]

It seems that environment-name now is a mandatory argument, even when using environment-file: false. That is a bit counter intuitive and a breaking change. I did not look into the details but how hard would be to maintain the old behavior and have the new changes? Maybe a default environment-name: nofile?

I believe this was introduced in #45 but I'm not sure. Trying to pin the GHA in all my CIs to get them back online.

[jonashaag]

It is mandatory if you use environment-file: false. What would you expect Mamba to use as environment name in that case?

Edit: Oh, or do you simply not care about the environment name?

[ocefpaf]

I do and it is set when I call micromamba. When we don't have an environment file you are usually handling things with the CLI, with less magic. This change makes the "just get me micromamba" option a bit more complex.

[jonashaag]

I see, good point. I think it is fine to not require a name if nothing is going to be installed.

[ocefpaf]

BTW the setup below is, IMO, a good setup b/c the lines in the yaml are almost dev install instructions:

- name: Setup Mamba
      uses: mamba-org/provision-with-micromamba@v11
      with:
        environment-file: false

    - name: Python ${{ matrix.python-version }}
      shell: bash -l {0}
      run: |
        micromamba create --name TEST python=${{ matrix.python-version }} --file requirements.txt --file requirements-dev.txt --channel conda-forge
        micromamba activate TEST

Sure I can add the env name there but it feels awkward b/c that option should exits only to override the name in the environment.yaml file. In fact one could argue that we can even rename the env from the CLI too when using an env file.

[hadim]

I got the same bug and agree with @ocefpaf. Our GA is at https://github.com/datamol-org/datamol/blob/a4b7cc16044ddb7d947e074a86ce5ffabf12eaec/.github/workflows/test.yml#L32

[jonashaag]

Can you test with the branch in the PR?