[WIP] Add correct CFLAGS to hostpython3 and -fPIC flag to avoid relocation error for few recipes

[rdbisme]

I think this fixes #2074 (and related), at least for me on macOS. I tried to build the KivyMD Kittchen Sink app for x86, armv7 and armv8-64, and went ok on my machine (macOS Catalina) using latest NDK 21.0.6113669 and also 19c.

PS: Most of the changes are related to automatic formatting using black.

[opacam]

Nice!!, but could you remove all the automatic black's formatting? 😁

We already have been discussed this at #1950 and doing some massive formating as you did I don't think that it's a good idea (specially now that we have some important changes, like python3 upgrade and the removal of python2 support)...It makes it hard to review and also it's easy to miss things...for instance...this has a tox error at here which blocks our tests.

[rdbisme]

Ok, I should have re-done all the edits disabling Black auto formatting (on my Vim 🤣 )

[rdbisme]

Ok, everything looks like it's working now. Only thing, I don't know how to fix the Travis error...

[AndreMiras]

The Travis error is probably a flaky test, I've restarted this specific job to see.
For future records this was the error:

    def test_blacklist():
        # First, get order without blacklist:
        build_order, python_modules, bs = get_recipe_order_and_bootstrap(
            ctx, ["python3", "kivy"], None
        )
        # Now, obtain again with blacklist:
        build_order_2, python_modules_2, bs_2 = get_recipe_order_and_bootstrap(
            ctx, ["python3", "kivy"], None, blacklist=["libffi"]
        )
        assert "libffi" not in build_order_2
>       assert set(build_order_2).union({"libffi"}) == set(build_order)
E       AssertionError: assert {'android', '...pyjnius', ...} == {'android', '...openssl', ...}
E         Extra items in the right set:
E         'genericndkbuild'
E         Use -v to get the full diff
tests/test_graph.py:96: AssertionError

And the PR looks good to me thanks 👍 I'll let @opacam have the final word for it

[rdbisme]

Ok, that test went well now.

[rdbisme]

Unfortunately, this does not fix the problem with pyconfig.h on macOS. I just realized that if I run the configure and then make command of the hostpython3 with a completely empty environment except for the CPPFLAG pointing to the pyconfig. It works. But then it breaks on Linux.

I also discovered this anomaly: https://bugs.python.org/issue40261

Can any of you with a mac try it out and check if it's getting the same?

[opacam]

@rubendibattista , there is an ongoing PR that fix the pyconfig.h issue (#2159)...could we deal with the -fPIC thing in here?

[rdbisme]

@opacam That's great. I guess those are two different problems. Maybe just land the PR about the pyconfig and I'll reabse and test againt the -fPIC flag. What do you think?

[opacam]

As you wish, but I think that you could do it whithout waiting if you want (I don't think that we have merge conflicts), whatever you decide is fine to me 😉