OpenBSD: Use our own implementation of wcsnlen()if not available

[daschuer]

This seems to be required to make our code compile with OpenBSD. See: https://mixxx.zulipchat.com/#narrow/stream/247620-development-help/topic/compiling.20mixxx.20on.20openbsd.20-current

It get also around the macro hell with MSVC/MingWG/wxWigets headers and avoids a redundant null check if actually the C++11 implementation is used.

Our implementation matches:
https://github.com/wxWidgets/wxWidgets/blob/fec8c06a53f5f2eb65e3328eadddd22334e7971f/include/wx/wxcrt.h#L183

[JoergAtGithub]

I don't think this is the right approach, because this for-loop replaces the much more efficient SIMD implementations of these standard functions, which would be used otherwise.

I think the real problem is, that strnlen and wcsnlen used in these wrappers are not standard functions. Instead wcsnlen_s and strnlen_s should be used. These are official C11 standard functions.

[daschuer]

I have no added a special case for OpenBSD only. Hopefully this solves the issue.
I have not used the proposed c++ function, because they are not available on all targets.
Using our own wrapper seems to more robust than fielding with
__STDC_LIB_EXT1__ __STDC_WANT_LIB_EXT1__ in a header that may cause unwanted side effects and test it on all our targets.

[JoergAtGithub]

I wonder if we've selected the C language version properly. Could you try set(CMAKE_C_STANDARD 11) or set(CMAKE_C_STANDARD 17) in CMakelists.txt

[JoergAtGithub]

It's C11, not C++11

[daschuer]

I wonder if we've selected the C language version properly. Could you try set(CMAKE_C_STANDARD 11) or set(CMAKE_C_STANDARD 17) in CMakelists.txt

This does not make any difference, because the C Standard is only set for c files, but we have only c++ files.
My gcc 9.4.0 is anyway reporting:

              GNU dialect of ISO C17.  This is the default for C code.
               GNU dialect of -std=c++14.  This is the default for C++ code.

So that should be fine. I have no OpenBSD, so I cannot test.

[Swiftb0y]

If we compile with -pedantic, gcc ensures we're using only ISO C/C++ instead of their GNU dialects. That should make it easier to ensure compatibility with *BSD...

[JoergAtGithub]

These bound checking _s functions of the C11 standard seems to be rather controversial: https://stackoverflow.com/questions/50724726/why-didnt-gcc-or-glibc-implement-s-functions/50724865#50724865

They are official part of the standard (Annex K), but it's optional to implement them.

The function wcsnlen (without _s) was never standardized.

[oneofvalts]

With this, I'm able to compile without getting an error on wcsnlen on OpenBSD.

[daschuer]

OK, Than this is ready for merge. @Swiftb0y can you have a final look?

[Swiftb0y]

LGTM

[Swiftb0y]

Feel free to squash or rebase or just merge yourself.