I’m already familiar with this, as it also stems from official documentation. However, many times I find myself being pushed by Ember to scatter responsibilities all around (e.g. in View, in Controller, and in Model).
To make it a bit more structured, I’m now adhering to the Xeroxish MVC. That is, a rendered View takes action through controller (target=controller) and only through controller (which makes the View virtually empty of actual methods).
The controller will then take action on the Model, which will cause the view to update by virtue of binding (which Ember is awesome at).
This completes a cycle of user input and refresh.
Given, it does make a ton of boilerplate and dull delegation, for example when a controller has nothing to do but just delegate to the model in turn. However from experience soon enough the need will arise and that same dull controller method will fill up with additional concerns that aren’t relevant to the model.
So, for me this is how I wrap my mind around the layers in Ember.
This kind of “rigidness” or rules, make a good training wheels as I go along, which makes me wonder about what are others doing.