get_in and its related lensing macros have the “default” or “built-in” behavior (for Map keys, the only thing they support by default) of being forgiving when you attempt to navigate further from a non-existent key. For example:

iex> get_in(%{}, [:a, :b])
nil

However, none of the “accessor” functions in Access (Access.at/1, Access.elem/1, Access.all/0)—which, as far as I understand, only exist to be used in this set of lensing macros—“follow the lead” of the built-in behavior of these functions, in terms of how they handle navigating from a nil parent.

iex> get_in(%{}, [:a, Access.at(0), :b])
** (RuntimeError) Access.at/1 expected a list, got: nil

Is there a reason for this difference?

Is it necessary—e.g. can the modifying lens macros (put_in et al) not work with non-strict accessors for other types, the way they work with non-strict accesses of map keys?

Or could there be safe/non-strict variants of these accessor-fun-returning functions that do work this way? (I’ve written such variants myself, but only for get_in.)