scoder

I added more visible warnings to the functions in 9d2ae85b2b5ae36deb68132152a56437e5cf9615, although I doubt that many people will read those, if they didn't even see the big grey boxes in the...

> Using `lst[i] = obj` seems to not be an option in this case, since it segfault either way. That is because you are creating a list in an invalid...

> Is there any way to just get the file from the installed Python distribution (in a sensible, reproducible way). I don't think so. They are not part of a...

>I'd probably put TypeStubGenerator.py through something like black to standardize the formatting. It's a lot better now but not perfect. We don't generally do this for existing files, but it's...

> attending the SciPy sprints and will be working on this Great. I'll push the final Cython release soon (it's actually tagged already), but given that this is an entirely...

Do you need to declare the dtype at all? I.e. do you need the buffer feature here? What do you do with a buffer of unknown dimensions? Could you provide...

Hmm, interesting. I'd say, most of the time, memory views are probably just passed around inside of one module or even live only in a single function. Even larger projects...

This is the C++ code that I get on my side: ```c++ /* "cpp_move.pyx":7 * def test(s): * cdef vector[string] v * v.push_back(move(s)) #

> add it to the main entry points for these facilities, ie the `cythonize` main function and to the `pyximport` module. Agreed for `pyximport`, which definitely depends on `distutils`. But...

> the variant is called `setuptools` and not `cythonize`. I think `cythonize` is a much better choice I disagree because `cythonize` is ambiguous. ~The `cythonize()` function shouldn't depend on setuptools~...