-
mapper now manages simple one-to-many relations, some things are hardcoded, but it's a start. Relations are lazy loaded, and for now it's necessary to call the .load() method on the loaded entity to populate the related fields.
-> we may need to autoload data on use? can be done later -> we need a way to chose a load strategy: eager, lazy ... not relevant for some storages, but sql will need it.
-
we may need to combine MemoryStorage with MemoryTable from test utilities, as their purpose is the same.
-> the proof of concept in bin/sandbox.py should be refactored into tests, for the stable parts.
- reworking mapper vs storage api. Storage should probably not know about mapper, and mapper should be responsible for mapping. Wow.
- writing some tests
- reworking (in the test, wip) the mapper interface to it's more decoupled and able to work with more than one underlying storage.