Using <M extends Model> for the repository functions?

[MartinX3]

I think it would be great to use as example

save(records: M[]): M[];
save(record: M): M;

instead of

save(records: Element[]): M[];
save(record: Element): M;

So my IDE could autocomplete all the object attributes I'm writing down to as example save one or multiple objects in the database.

At the moment I do useRepo(Bla, store).save([{...}, {...}, {...}] as Bla[]) to have auto completion and errors if I miss a non optional field.

[CodeDredd]

I think that's not a good idea, because maybe you want to pass an property to the save function which is not belonging to the model, but you want to react on it with a hook. If i type check the save function then this is no longer possible.
Another point are relations. It would be really hard to type check an big dataset with relations.

[CodeDredd]

Do you have an suggestion to make sure I don't miss mandatory fields and maybe get an autocompletion?

Use the hooks for that. E.g. the creating hook and return false if one mandtory prop is missing with an error message. That way it's not saved and you know what to look at.

And what would be a use case to save fields which I didn't specify in the entity model?

It's not about saving. The props could be used e.g. in the creating hook where you check maybe a specific feature flag which triggers something but is not important to be saved.

[MartinX3]

Ahhhhh, thank you!

[CodeDredd]

This brings me actually to the idea to introduce a required flag for fields which if not set are throwing an error and canceling the save.

[MartinX3]

I thought non optional/null fields are mandatory by default.

[CodeDredd]

true....but they are only throwing a warning. I think a feature flag is needed if you want to throw this as error and canceling the action.