Compare with pydantic settings

[MartenBE]

I am looking for a good solution to parse yaml as config files for a python CLI tool. I both found this library and pydantic_settings, but I don't understand how the compare or differ. Could you elaborate on that and give me some advice?

[NowanIlfideme]

Hi, sure.

This library is meant to parse things from YAML and save them to YAML, since Pydantic is based primarily on JSON compatibility. It existed before Pydantic v2, hence a surprising number of projects are using it: ~20k per weekday.

pydantic-settings is a first-party library (developed by the Pydantic folks). It's meant primarily as a way to work with environment variables, also with command-line parameters or from "dotenv" (.env) files.

Actually, reviewing now, I can see they have a YamlConfigSettingsSource as well as JSON, TOML, pyproject.toml - I will have to check it out. If it's robust enough for use-cases, I might dd docs/deprecation warnings, urging folks to switch, and how to migrate; then stop maintaining/developing this library. I'd much rather this be handled uniformly rather than as a third-party-developed package. 😉

I was also considering adding support for OmegaConf as a YAML resolver, possibly later primary one, since they have templating support.

[NowanIlfideme]

Looking further, it seems that pydantic_settings isn't a full replacement. It's meant as a source of YAML for settings, rather than also supporting writing. If your use-case is reading configuration for a CLI app, then I'd go with pydantic_settings. If you want to read/write YAML files with Pydantic models, it seems like pydantic-yaml would be the way to go.