Cannot compile in Visual Studio 2015

[CurtisHx]

Cannot compile in Visual Studio 2015

[CurtisHx]

It looks like it's a difference in the way GCC and Visual Studio treat __VA_ARGS__. GCC expands __VA_ARGS___ before passing it into the macro. Visual Studio expands __VA_ARGS__ after passing it.

See this answer on Stack Overflow: http://stackoverflow.com/a/5530998

A fix is to force the macro expansion by wrapping macros taking in __VA_ARGS__ with this macro:

#define EXPAND (x) x

For example, the FAKE_VOID_FUNC call chain in fff.h goes from

#define FAKE_VOID_FUNC(...)     FUNC_VOID_(PP_NARG_MINUS1(__VA_ARGS__), __VA_ARGS__)

#define FUNC_VOID_(N,...)     FUNC_VOID_N(N,__VA_ARGS__)

#define FUNC_VOID_N(N,...)     FAKE_VOID_FUNC ## N(__VA_ARGS__)

to

#define FAKE_VOID_FUNC(...)     EXPAND(FUNC_VOID_(PP_NARG_MINUS1(__VA_ARGS__), __VA_ARGS__))

#define FUNC_VOID_(N,...)     EXPAND(FUNC_VOID_N(N,__VA_ARGS__))

#define FUNC_VOID_N(N,...)     EXPAND(FAKE_VOID_FUNC ## N(__VA_ARGS__))

In addition, the PP_NARG_MINUS1 and PP_NARG_MINUS1_ macros need to be wrapped similarly.

I'm 99% confident this fix will work with GCC as well as VC++. I'm just not sure how to go about making the change in the ruby generation file.

[CurtisHx]

Tracking it down further (I don't have a solution yet...), it has to do with the way that the VC++ preprocessor expands (or lack thereof) of __VA_ARGS__. Of course Microsoft would have to do it differently from everyone else...

[shebaw]

You can explicitly use the numbered functions as a workaround.
Example:
FAKE_VALUE_FUNC1(void *, malloc, size_t)

[VincentDary]

I work with gcc on linux and cl.exe on windows, the EXPAND trick work well for M$. Thank's

[meekrosoft]

Thanks for the contribution!