Microsoft VC Compiler Incompatibility about minfft HOT 9 CLOSED

Thank you for reporting the issue.

It's a pity MSVC behaves like that. About support - well, I would rather prefer for my library to adhere to standards, not to specific compilers...

In the meantime, please try to change a definition in minfft.h from
typedef double complex minfft_cmpl;
to
typedef _Dcomplex minfft_cmpl;
and see if it helps.

from minfft.

@beecksche did this solve the issue?

from minfft.

I've started to implement it, but hadn't time to test it yet.

Unfortunately it's not done with only replace typedef double complex minfft_cmpl with typedef _Dcomplex minfft_cmpl. I had to change some parts of the code aswell. I hope I can send a pull request end of the week with some pre-processor directives for MSVC compiler.

from minfft.

@aimukhin, Now, the code can be complied additonally in C++ with MSVC (2017 & 2019, 32bit & 64bit).

But I haven't tested it! You wrote, that you used the benchFFT. Do you have a code snippet, or a readme how to use benchFFT with minfft?

from minfft.

@beecksche minfft is included in the current master benchFFT

from minfft.

@beecksche I just pushed two commits that fix the library code for MSVC. Hopefully, changing double complex to _Dcomplex should be the only thing you have to do. Please try it and let me know.

from minfft.

That's awesome! Works perfectly, and a much more elegant way than mine.

In the header I also had to add

#ifdef __cplusplus
extern "C" {
#endif
...
#ifdef __cplusplus
}
#endif

for C++ compiling.

from minfft.

@beecksche I just pushed commits to further improve compatibility. If C99 complex is not available (as with MSVC), then minfft_cmpl is an array of two reals, which is binary compatible with _Dcomplex and complex<double>. You only have to cast pointers:

_Dcomplex z[N],w[N];
or
complex<double> z[N],w[N];
...
minfft_dft((minfft_cmpl*)z,(minfft_cmpl*)w,a);

Also I added extern "C" wrapper to the header file.
Hope you can now use the library without any modifications.

from minfft.

@aimukhin Awesome, worked perfectly.

from minfft.