Windows: inconsistent redefinition of _aligned_malloc about corrade

[mosra]

Yep, this bothers me as well 😅

Oh, it's __declspec(restrict)? Then it should be fine to put that there. I can't read, I thought it was __declspec(dllimport) and such, meaning I'd have to do detection whether static or dynamic runtime is used, using the right flag, inevitably running into more problems...

I have a rule to never include anything from Windows in any public header, because then, if your project doesn't define all the NOMINMAX, LEAN_MEAN_LEAN_EXTRA_LEAN_MEAN etc macro, you get stuff like #define interface struct and it's not great. X11 includes are the same, also banned from public headers. An alternative solution I was considering was to add Memory.cpp and chuck the #include <malloc.h> in there, but since that's one more indirection and exported symbol I'm not so sure about the tradeoff.

Actually, let me first check what <malloc.h> has on Windows. Since I don't see a "sanitized" version of this header in https://github.com/Leandros/WindowsHModular, it's probably not so bad after all...

[mosra]

Fixed in d3f4a0c -- the Windows <malloc.h> doesn't seem to have that much cruft, after all.

[pezcode]

Thank you 🙇