ViewModel: Function fields vs function literals?

[hacker1024]

In many examples (including the TODO architecture sample), ViewModels are created like so:

class ViewModel {
  final void Function() myFunction;
  final String myProperty;

  ViewModel({
    required this.myFunction,
    required this.myProperty,
  });

  factory ViewModel.fromStore(Store<AppState> store) {
    return ViewModel(
      myFunction: () => store.dispatch(const MyFunctionAction()),
      myProperty: store.state.myProperty,
    );
  }
}

This (IMO) has the following problems

• myFunction may be accidentally used in a == or hashCode implementation, which would be incorrect as the closure would be non-identical with every new ViewModel
• More RAM usage (maybe?)

Why is that the recommended approach, compared to something like this?

class ViewModel {
  final Store<AppState> _store;

  ViewModel(this._store) : myProperty = _store.state.myProperty;

  final String myProperty;

  void myFunction() => _store.dispatch(const MyFunctionAction());
}

[brianegan]

Agreed -- your approach is a bit better for the reasons you state!

I was originally playing around with some ideas of creating the ViewModel more in the build function, where it's nice to be able to define those callback functions.

However, as I began to realize it's actually nicer to create the view Model in a static factory / normal class constructor, it now makes less sense.