mv-python

@thedrow this is great result. What test did you run? I would like to repeat it also against cythonized version just for curiosity. Is it able to build against windows?...

@thedrow, see my comments in the patch. The problem is that when you are using publisher confirm, the publisher checks before publishing if broker sent a message to him. This...

Unfortunately, I am not able to replicate it: ``` >python.exe Python 3.8.0 (tags/v3.8.0:fa919fd, Oct 14 2019, 19:37:50) [MSC v.1916 64 bit (AMD64)] on win32 Type "help", "copyright", "credits" or "license"...

> For reasons I don't fully understand, I have to use noexcept to get the program to compile: I think your example makes sense to me to fail since in...

BTW this PR seems OK for me except comment by @scoder.

> This is not how this should work. "extern" functions shouldn't raise exceptions, unless explicitly marked as such. They are just too unlikely to raise in the real world. Right...

One easy improvement is that full CI is skiped for documentation-only changes. E.g. we can specify to run tests only when files in `Cython/*` are changed [1]. E.g. pylint is...

I have investigated also the cause of #4552. It seems that `analyse_declarations()` is called for package `PKG` following two ModuleScopes: * `PKG` * `PKG.__init__` In our case for package `pkg.sub`...

> I could be wrong (I've only had a quick look through) but: ModuleScope has an is_package attribute. Yes I saw that. But it is set only for `__init__` ModuleScopes...

This is the first attempt to fix the issue. On my local computer it fixes the TC, I hope I did not introduce any regression. I have tried to set...