add config.schema.json

[cjyetman]

No description provided.

[cjyetman]

@AlexAxthelm this is still in draft mode, assuming you may have some suggestions. Also still need to figure out appropriate values for the "$id", "title", and "description" keys... and configure the GH Action appropriately and make sure it works as expected.

[AlexAxthelm]

Exciting! I'll take a closer look later!

[AlexAxthelm]

@cjyetman Noting a couple of things here from the docs used in the action:

First: it doesn't look like we're actually validating against the schema you wrote. From the docs:

Input	Required?	Default	Description
yaml_schema	false	""	The full path to the YAML schema file (e.g. ./schemas/schema.yaml) - Default is "" which doesn't enforce a strict schema

So right now it's just doing a check on "is this a valid yaml file", not "does this yaml file match this schema".

Second: I don't think we can make a good schema that will cover the inheritance that we're counting on with config::get().

The Schema should model a fully-defined object, but as we have the files written (2023Q4_azure inherits from 2023Q4 inherits from docker inherits from default), none of those objects will necessarily have all the correct keys.

Here's an article about modeling inheritance with JSON schema: https://json-schema.org/blog/posts/modelling-inheritance but the short story is that it isn't as intuitive as config makes it.

I think this absolutely makes sense for us to chase down, but it might involve bigger changes to how we handle/import configs than just doing the check.

[AlexAxthelm]

One "workaround" option would be that for each of the configs, we read them in with config::get() (to deal with inheritance), and write them to file before checking them against schema.

[cjyetman]

Yeah, I had a feeling this validator action isn't going to work with the inheritance easily. I'm definitely not wedded to it, just thought it would be cool if it worked.

[cjyetman]

gonna remove the validator action because I don't think there's any reasonable way of making it work as expected/desired

[github-actions]

Docker build status

Commit time	Git sha	Image
2024-04-30T06:17:54Z	b293b74	ghcr.io/rmi-pacta/workflow.data.preparation:pr-167

History

Commit time	Git sha	Image
2024-04-30T06:17:54Z	b293b74	ghcr.io/rmi-pacta/workflow.data.preparation:pr-167

History JSON

`[{"commit_time":"2024-04-30T06:17:54Z","git_sha":"b293b74f996b1c2c433fd086794e77cc92fcd2f5","image":"ghcr.io/rmi-pacta/workflow.data.preparation:pr-167"}]`

[cjyetman]

While this was a cool experiment, I don't know if this would ever get used and it has already fallen out of date, which suggests to me this will be more of a maintenance burden than useful, so I'm gonna close it.