On Fri, Apr 22, 2016 at 4:56 AM, Christian Ullrich <chris@chrullrich.net> wrote:
> * From: Andrew Dunstan [mailto:andrew@dunslane.net]
>
>> 4. The compiler complains about one of Microsoft's own header files -
>> essentially it dislikes the=is construct:
>>
>> typedef enum { ... };
>>
>> It would be nice to make it shut up about it.
>
> I doubt that's possible; the declaration *is* wrong after all. We could turn off the warning:
>
> #pragma warning(push)
> #pragma warning(disable : 1234, or whatever the number is)
> #include <whatever.h>
> #pragma warning(pop)
Well, yes.. Even if that's not pretty that would not be the first one
caused by a VS header bug (float.c)..
>> 5. It also complains about us casting a pid_t to a HANDLE in
>> pg_basebackup.c. Not sure what to do about that.
>
> The thing that's being cast is not a PID, but a HANDLE to a process. pid_t is a typedef for int (in port/win32.h),
thereforeis always 32 bits, while HANDLE is actually void*. However, Microsoft guarantees that kernel32 HANDLEs (this
includesthose to threads and processes) fit into 32 bits on AMD64.
>
> Source: https://msdn.microsoft.com/en-us/library/windows/desktop/ee872017(v=vs.85).aspx, third bullet point.
>
> So we can simply silence the warning by casting explicitly.
Yes, when casting things this way I think that a comment would be fine
in the code. We could do that as separate patches actually.
--
Michael