Few API questions about model HOT 4 CLOSED

@sidonath Thanks for asking these questions.

Collections are mapping low level details, they shouldn't be used directly. Repositories and queries are the boundary for the persistence mechanisms. In other words, your code shouldn't be aware of collections.

What's the goal that you want to accomplish? Please point to a concrete example, and I'll be happy to help in the specific case.

The reason for keeping those attributes assignment separated is the intrinsic nature of a mapper. The simplified examples and your greenfield project are using one entity per one database table (collection). This architecture shines on more complex cases. Think of a Customer entity that is composed by complex data such as Address, CreditCard. A mapper is able to persist/fetch those entities from/to separated database tables, in a way that is transparent to the domain model.

Martin Fowler in his EAA book admits that Data Mapper is only recommended for complex cases, Active Record (the pattern), should be preferred otherwise.

For this reason I'm thinking to an "auto mapper" mode for Model: it will be able to automatically map the schema and generate attributes for entities. It will be recommended for prototypes and for early stages of an app, and should be disabled when it "hurts".

from model.

What's the goal that you want to accomplish? Please point to a concrete example, and I'll be happy to help in the specific case.

I'm trying to extract the boilerplate for setting up the Lotus::Model objects. Basically, playing with the job that Lotus gem is going to do. Take a look at this file, which is a modified version of lotus.rb from your gist:

https://github.com/sidonath/room-reservation/blob/3177a349f585d168b058fe5c6013454dfcaad876/lotus.rb#L39

Think of a Customer entity that is composed by complex data such as Address, CreditCard. A mapper is able to persist/fetch those entities from/to separated database tables, in a way that is transparent to the domain model.

I haven't considered that, but it sounds awesome! I guess that's planned for future, since the current code — as far as I can see — assumes that the relationship between a collection and an entity is 1-to-1?

For this reason I'm thinking to an "auto mapper" mode for Model: it will be able to automatically map the schema and generate attributes for entities. It will be recommended for prototypes and for early stages of an app, and should be disabled when it "hurts".

Great! I was pretty certain it was planned 😄

from model.

@sidonath Have a look at this commit ec395d8 it introduces a getter for the collections.

Yes, in a future version, Model will be able to handle those complex persistence cases.

from model.

@jodosha awesome! Thanks for clearing up everything 😄

from model.