- [X] Have you followed the [guidelines for contributing](https://github.com/canonical/snapcraft/blob/main/CONTRIBUTING.md)? - [X] Have you signed the [CLA](http://www.ubuntu.com/legal/contributors/)? - [x] Have you successfully run `make lint`? - [X] Have you successfully run `make test`? --- I created the new inheritance based Project validation, with the simplified scope discussed in the clinics. I took the charmcraft project as inspiration for the schema generation, and validation, used its pipeline as my base for the schema.yaml pipeline to assure that the schema file will always match the current state of the code. In the charmcraft project, there is a validation of the schema, made using files from the smoketests directory. We do not have a smoketests directory in the snapcraft project. Is there a place where there are files ready for this, should I add these files in a specific directory, or is it out of scope? For now that job is blocked so it won't be run. In the `Makefile`, I only added the task to generate the schema. It is to `new_snapcraft.json` so the reviewer can check if the generated json is in the right format, having the old one as a comparison. If the schema generation is approved, I will make a commit changing to `snapcraft.json` as it should be. In `snapcraft.models.__init__.py`, I only exported the new models, so they can be imported in the tests. In `snapcraft.models.project.py` I created the project models, and the validation logic, and moved some validation from the monolithic model to the new models. Also, I overrode the unmarshal function to validate the data through the TypeAdapter of `SnapcraftProject`. I had to use Discriminator functions, instead of field discriminators, because as this was an initial, with only three supported bases and two build bases, there should be a default fallback for the ones that are neither of them, but I used the enums, to generate them based on the Enum, as discussed. In `conftest.py` I made a small change in the fixture `setup_project` to use unmarshall instead of model_validate, so the correct submodel will be used, as specific validations are made by them. In `tests/unit/models/test_projects.py` I added a test for the unmarshaller, inside of the `TestProjectValidation` class, and changed a line that checked for "core22" base to check an instance of Core22Project, so mypy would not complain. In `tests/unit/services/test_buildplan.py` I just changed a Project.model_validate to Project.unmarshall to use the correct submodel. In `tools/schema.py` I based myself in the `tools/schema.py` of the charmcraft project. This PR is still not ready to be merged, since I need some guidance in relation to the validate-schema job, and also if the generated schema is compliant with the expectations. I would also like to know if there is anything missing, or anything that I got wrong so I can fix it. I did not change the docs, to document my change, because I was not sure if it is in the scope of this issue. I am also sorry for the delay in this delivery, but I will be alert, waiting for the comments, and review. Fixes #5543