Support a yaml-based format for Standard Names dictionary

[mkavulich]

Many applications support and/or rely on YAML format for various input files. YAML has the advantage of being widely supported in Python and other applications, and is more human-readable than the current XML dictionary. Some options for incorporating this as an option for the Standard Names dictionary are:

• Replace the existing standard_names.xml with a YAML format for the main dictionary file
  • This could be problematic for back-compatibility
• Add standard_names.yaml as a third location where the dictionary is hosted, automatically created from standard_names.xml same as Metadata-standard-names.md
  • Yet another file to maintain is not ideal
• Replace the existing human-readable Metadata-standard-names.md
  • This could be a good compromise, though YAML is not quite as human-readable as the rendered markdown

[gold2718]

YAML has the advantage of being widely supported in Python

While I am a fan of YAML, I do want to point out that it is currently less supported than XML in Python. Python has various XML tools as part of its standard library while YAML is not. There are different outside YAML libraries with different strengths, weaknesses, and bugs. If we intend to support YAML as an input format (as opposed to your generated option), I feel we should define an allowed syntax and implement a syntax checker to enforce our rules. We will also need to publish and maintain instructions for installing and using YAML libraries on a representative variety of platforms.

[gold2718]

• Yet another file to maintain is not ideal

Could we move generation of the Markdown and/or YAML formats to a GitHub action? That would really streamline development although just adding a YAML output to the current script (which was originally written to support multiple output formats) would not increase the workload.

[climbfuji]

• Yet another file to maintain is not ideal

Could we move generation of the Markdown and/or YAML formats to a GitHub action? That would really streamline development although just adding a YAML output to the current script (which was originally written to support multiple output formats) would not increase the workload.

That's a good idea. And then produce .md and .yaml at the same time.

I can assign this task to me unless someone else is eager to do so.

[danrosen25]

@danrosen25 - tagging myself for tracking

[mkavulich]

@gold2718 @climbfuji There currently is a GitHub action running write_standard_name_table.py, but it simply checks that the XML and .md files are equivalent (check-rerendered-markdown in https://github.com/ESCOMP/ESMStandardNames/blob/main/.github/workflows/pull_request_ci.yml). Would the idea with creating the YAML through GitHub actions be that we would move away from making users run the write_standard_name_table.py script all together? Or simply add an additional check that the updated YAML has also been created?

[gold2718]

@mkavulich, I am in favor of having the action produce both files. I'm sure folks can get used to only having to edit one file :)

[climbfuji]

@mkavulich If you want to work on this, then please let me know. I was about to start.

[mkavulich]

@climbfuji you go for it, I'm on PTO for the next few days so I won't have time for a bit.

[climbfuji]

PR: #117