Unlike the other collections ABCs, MappingView and its subclasses (Keys|Items|Values)View don't define __slots__, so users inheriting them to take advantage of the mix-in methods get a __dict__ and __weakref__, even if they try to avoid it by declaring their own __slots__. This is sub-optimal (I don't think any library class should leave __slots__ undefined, but it's particularly bad when they're intended to be used a base classes).I've attached a patch that defines __slots__ for all of them; MappingView explicitly declares its _mapping slot, and the rest declare no slots at all.Only negative I can think of is that if the user provides their own __init__, doesn't call the super().__init__, and uses a different name than _mapping for whatever data structure they actually use, then there will be a pointer reserved for _mapping that never gets set. That said, if they don't define _mapping, none of the mix-in methods work anyway, so they may as well not inherit directly, and instead just use *View.register to become a virtual subclass.

Hmm... First patch isn't generating a review diff (I'm guessing because my VM's time sync went farkakte). Trying again.

I also wrote a slightly more thorough patch that simplifies some of the method implementations (largely just replacing a bunch of for/if/return True/False else return False/True with any/all calls against a generator expression).The one behavior difference is that I dramatically simplified Sequence's __iter__; the original just kept indexing until it got IndexError, the replacement just runs len(self) times, on the theory that a self-mutating Sequence should write its own __iter__, and we shouldn't jump through hoops to accommodate unsupported behavior like mutating a Sequence during iteration.All tests pass under both patches.

At first glance, this seems like a reasonable request.I leave it open for comments for a while before proceeding.

New changeset2c6a231e2c1e by Raymond Hettinger in branch 'default':Issue#21421:  Add __slots__ to the MappingViews ABCs.http://hg.python.org/cpython/rev/2c6a231e2c1e

I've applied the __slots__ patch.  Thank you for submitting it.Am leaving the rest of the ABCs code as-is.  The current form may be a bit wordy but it is clean, fast, and easy to trace through a debugger.  The expanded forms were chosen for a reason.

Cool. Thanks for the feedback. I split it into two patches for a reason; I wasn't wedded to the simplification.