Many redundant methods in subclasses

[rwbarton]

For example, dict is a subclass of MutableMapping and hence also Mapping and other classes, and almost all of dict's methods are actually provided by one of those superclasses with the same signature. Is there a good reason to repeat the declarations of these methods in dict? (Perhaps they predate dict's superclasses?) I tried removing get and there seemed to be no ill effects.

One possible reason: it's convenient to see the types of all of dict's methods in one place. But it also creates more work when changing these types, as in #223. For being able to see all the methods of dict at once, we could have mypy automatically generate documentation (or perhaps a python documentation tool already understands function annotations).

[gvanrossum]

I think it's mostly a historical accident. I would generally discourage people from reading the stubs as a form of documentation, they're really not meant for that.

[srittau]

I am going to close this. While those redundant definitions are not harmful, we gladly accept patches that will remove these redundancies.