As with many projects that bubble to the top of HN, I wish there was more justification of what problem this is trying to solve. Of course "just for fun" is a totally valid reason, but this seems to go beyond that and it's not clear to me why this is a better solution than existing languages.
Lately I've been looking at the generation of the YAML files. I'm not a fan. PKL was released and seemed promising, but support is limited (and I'd say after trying to build some stuff - tooling isn't that great).
Internet searches hadn't brought me anything.
Does anyone has some success stories with working with multiple YAML documents or is this always ending up in "just hack bunch of scripts, because alternatives are worse" ?
We define all our YAML config in typescript and just serialize it out. It’s a CI error to update the config scripts but not commit the newly generated YAMLs so nothing gets out of sync, and it’s usually only a few minutes work to write a typedef for a new config file type. Highly recommended.
Edit, not similar to what (PuppetLabs) Hiera does. But I've been using Hiera for years, first with puppet, then with terraform. Not to program, but simply to store config in a hierarchical store. There is crossover. But this seems to go deeper into the type of stuff that people don't want to use yaml for. It's interesting though, so I will check it out.
Would love if there were more parsers for nix lang to turn them into language specific datastructures. E.g. rust serde for nix. Nix covers much of the same features as YAML but is intentionally Turing complete.
