Using the same attribute twice does not seem to work even with dump_only=True on a Nested field

[aphilas]

I intend to use the same schema for loading and dumping. An example structure for this use case is: Author -< Book

Loading

AuthorSchema field: books = RelatedList(Related(), attribute='books') AuthorSchema().load({ 'books': [ 1, 2 ] }) # {'books': [<Book>, <Book> ] } ✅

Dumping

AuthorSchema field: books = Nested('BookSchema') AuthorSchema().dump(Author.query.get(1)) # {'books': {'id': 1, 'title': 'Lorem ipsum' }} ✅

Both

Having books = RelatedList(Related(), attribute='books') and books_list = Nested('BookSchema', dump_only=True) as explained in https://github.com/marshmallow-code/marshmallow/issues/1038, works for loading but not for dumping. Should the field have the same name as the attribute? If so that would make it impossible for both to exist simultaneously.

However using Pluck as in https://github.com/marshmallow-code/marshmallow/issues/1038 gives the desired effect, only it would be tedious to replicate for all fields.

I might have an incomplete idea of how marshmallow/marshmallow-sqlalchemy works, so please feel free to elaborate, or to ask for further clarification on my specific use case.

Addendum: I've noticed that in the case where you'd want to load a new Author as well as new nested Books e.g. AuthorSchema().load({ books: [{ 'title': 'My new book' }]}) (as show in this SO questions) you need the Nested field.

ETA: change example to more popular Author/Books model