InMemoryDao

[lhaze]

• represents contents of a file
• uses a loader, eg: pca.utils.serialization.CustomLoader or a json.loads/dumps
• often used to store configuration (data less structured and more slowly changeable than contents DB and yet still not the codebase)
• can use read-only mode
• can be whole loaded into memory on application startup or lazy loaded on demand

[lhaze]

As issue #22 makes usage of files as DAO possible & quite simple, it introduces a new dependency: tinydb. But still, there might be a place for own file DAO:

• As a framework, PCA should minimize the number of dependencies.
• There is a need for read-only DAO for things that change slower than the contents of the DBs, and yet they are changable - like settings, but in a shape of domain entities.
• It should be possible to load the entities from different types of files: yaml, json or ini.

[lhaze]

During implementation, I've decided that loading contents is not the responsibility of the DAO. Loading DTOs (dicts) from some kind of a file is the case for pca.utils.serialization, and pushing them into DAO should be done at the time of the DI registration. Whatever type of DAO it is -- not necessarily InMemoryDao.

So IMO the work here is done.