* Use Tuple field in DecimalTuple
* Remove unnecessary base classes from Decimal
* Decimal.__init__ -> __new__
* Decimal.__ne__ is not defined in Python 3
* Add Decimal.as_integer_ratio()
* Annotate DecimalException.handle()
* Correct types of Decimal method arguments
* Add missing arguments and optional markers to Decimal
* Add missing arguments to Context
* Remove spurious int from Unions with float
* Remove Context.__setattr__()
* Fix return types of Context methods
* pkg_resources.Distribution: Fix self arguments
* docutils: Remove spurious ellipses
* Add self argument to markupsafe.Markup.format()
While the actual signature is also "def format(*args, **kwargs)", the
implementation extracts "self" from "args" as first step.
* werkzeug: Remove methods not present during runtime
UpdateDictMixin.calls_update() and EnvironBuilder.form_property()
get deleted while constructing the respective class.
* requests: Add missing self argument
These type annotations are unnecessary and seem to confuse the type system.
* Remove annotation from PlistFormat enum members
Same rationale as python/typeshed#2314; same adverse effects observed.
Click already had stubs but not for the `testing` submodule.
I used stubgen and then narrowed the types when it was easy. I left
`Any` where more work was necessary.
Also, mark AST._attributes and _fields as class vars.
`lineno` and `col_offset` were previously defined on a few sub-classes of `AST`, e.g. `expr`, even though https://docs.python.org/3/library/ast.html explicitly states that `AST` has these two attributes. These attributes are only present if they were supplied as arguments to the constructor, but the same is true for the subclasses.
To backport `os.path.commonpath` in mypy I needed to use genericpath. It seems unchanged since 3.4, and the `same*` functions were added in 3.4. (checked via comparing the `__all__`s of the source in 2.7, 3.4, and 3.7.)
* Drop support for Python 3.3
* Merge Python 2 and 3 shutil
* Marked some arguments optional
* Changed callback return type from None to Any for more flexibility
I found some code that called DEFINE_integer() and DEFINE_float() with default set to None. Perusing the code that seems to be a totally valid usage. I am not super familiar with gflags but I believe this PR consistently changes all those default values to Optional[].
