Bug: openjd check does not handle duplicate keys

[AWS-Samuel]

Describe the bug

Given a job template, which has duplicate keys. openjd check will consider the job template as valid, even though it's not valid yaml.

Expected Behaviour

The template will not be considered valid, due to having duplicate keys

Current Behaviour

The template will be considered valid, even with duplicate keys.

Reproduction Steps

Given the template:

specificationVersion: 'jobtemplate-2023-09'
specificationVersion: 'jobtemplate-2023-09'
name: Foo
name: Foo
steps:
- name: Bar
  script:
    actions:
      onRun:
        command: Baz
        command: Baz
      onRun:
        command: Baz
        command: Baz
    actions:
      onRun:
        command: Baz
        command: Baz
      onRun:
        command: Baz
        command: Baz

> openjd check template.yaml
Template at 'template.yaml' passes validation checks.

[crowecawcaw]

Python's YAML safe loader overrides duplicates, and the underlying library uses the safeloader here: https://github.com/OpenJobDescription/openjd-model-for-python/blob/mainline/src/openjd/model/_parse.py#L40-L46

Suggest closing as "won't fix".

[epmog]

Python's YAML safe loader overrides duplicates, and the underlying library uses the safeloader here: https://github.com/OpenJobDescription/openjd-model-for-python/blob/mainline/src/openjd/model/_parse.py#L40-L46

Suggest closing as "won't fix".

Mmmm, that seems like an implementation detail rather than desired/defined behaviour. This makes me believe we need to do something here. Either update the spec to indicate duplicates do overrides, or update the spec and the library to consider these not valid job templates.

Looking at it, I'd consider it to be non-valid, since it's effectively dead code in a template and it won't ever be used, thus making it very misleading for users