Avoid const-ness compiler warning in __Pyx_PyObject_AsWritableString

[mbuesch]

This avoids the compiler warning
warning: cast discards ‘const’ qualifier from pointer target type [-Wcast-qual]
by casting the pointer to uintptr_t before removing the const-ness.

The warning was introduced in 6cd70f0

Also merge the two places that define stdint types.

[scoder]

I couldn't find this as a standard define anywhere, so I guess it's a name that you made up. In that case, we should prefix it with __PYX_, which is Cython's internal standard prefix.

[scoder]

(Although I wonder if this is needed at all – this code section can only appear once in each generated C file.)

[mbuesch]

_UINTPTR_T_DEFINED is what my Visual Studio 2017 Community uses as guard around its uintptr_t typedef. So I think we should at least keep the #ifndef _UINTPTR_T_DEFINED. But whether we should #define _UINTPTR_T_DEFINED, I'm not so sure. It could avoid compilation issues, if the system header is included later on (by user code).
What do you think?

[scoder]

If I read this page correctly, then MSVC defines uintptr_t in stddef.h. Maybe we should include that? (And then use the same guard as whatever they use?)

[scoder]

BTW, if MSVC defines uintptr_t in stddef.h, then I think we should have a test that includes stddef.h and exercises the AsWritableString macro code, to make sure that both don't interfere.

[mbuesch]

On my version of Visual Studio 2017 Community (which should be the latest one) uintptr_t is defined in vadefs.h.
I did not find another typedef in any other header file.
It's very similar to this file:
https://github.com/icestudent/vc-19-changes/blob/master/vadefs.h

There are quite a few libraries that also use _UINTPTR_T_DEFINED in the way I used it.
https://www.google.de/search?q=%22_UINTPTR_T_DEFINED%22
See libSDL for instance:
https://hg.libsdl.org/SDL/file/9e13f3286831/include/SDL_config_windows.h#l42
This one also wraps it with
#if !defined(STDINT_H) && (!defined(HAVE_STDINT_H) || !_HAVE_STDINT_H)

This seems to be a real mess.
What if we do not typedef uintptr_t and use our own name instead?
Something like this:

#ifdef _MSC_VER
    #ifndef _MSC_STDINT_H_
        #if _MSC_VER < 1300
           typedef unsigned char     uint8_t;
           typedef unsigned short    uint16_t;
           typedef unsigned int      uint32_t;
        #else
           typedef unsigned __int8   uint8_t;
           typedef unsigned __int16  uint16_t;
           typedef unsigned __int32  uint32_t;
        #endif
    #endif
    #if _MSC_VER < 1300
        #ifdef _WIN64
            typedef unsigned long long  __pyx_uintptr_t;
        #else
            typedef unsigned int        __pyx_uintptr_t;
        #endif
    #else
        #ifdef _WIN64
            typedef unsigned __int64    __pyx_uintptr_t;
        #else
            typedef unsigned __int32    __pyx_uintptr_t;
        #endif
    #endif
#else
   #include <stdint.h>
   typedef uintptr_t __pyx_uintptr_t;
#endif

...

#define __Pyx_PyObject_AsWritableString(s) ((char*)(__pyx_uintptr_t) __Pyx_PyObject_AsString(s))
...

[mbuesch]

Should I redo the pull request with the naming change?

[scoder]

Yeah, I think that looks safer. Doesn't help users who need this, but also doesn't interfere with whatever solution they find for themselves (which they probably need to do anyway).

[mbuesch]

I updated the commit. Please let me know if there's anything wrong with it.

[scoder]

Thanks for cleaning this up.

[mbuesch]

The reasoning behind this change was:
With this change (and the previous one that has already been merged) I can compile my project with -Wcast-qual without any warnings from Cython's internal code.
That warning option helped me to spot an actual incorrect cast in my user code. So I think it's useful for Cython to be compatible with that warning option.

[scoder]

Yep, that's the intended indentation. :)