`_VSTD::isnan` – "error: expected unqualified-id" on Clang

[MacDada]

Hey, it's me with C++ again ;)

I've just changed the stock Apple's Clang into the "generic" Clang from Homebrew. Everything works but one thing – some of my unity tests stopped working.

➜  DnWiFiDoorLock git:(master) ✗ clang -v
Homebrew clang version 15.0.7
Target: x86_64-apple-darwin21.6.0
Thread model: posix
InstalledDir: /usr/local/opt/llvm/bin

➜  DnWiFiDoorLock git:(master) ✗ pio test
Verbosity level can be increased via `-v, -vv, or -vvv` option
Collected 23 tests

Processing native/DnAppLogger/test_Logger in native environment
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Building...
Library Manager: Installing throwtheswitch/Unity @ ^2.5.2
Unpacking  [####################################]  100%
Library Manager: Unity@2.5.2 has been installed!
In file included from test/native/DnAppLogger/test_Logger/LoggerTest.cpp:3:
In file included from lib/DnAppLogger/src/DnApp/Logger/Logger.h:3:
In file included from /usr/local/opt/llvm/bin/../include/c++/v1/memory:853:
In file included from /usr/local/opt/llvm/bin/../include/c++/v1/__memory/compressed_pair.h:16:
In file included from /usr/local/opt/llvm/bin/../include/c++/v1/tuple:227:
In file included from /usr/local/opt/llvm/bin/../include/c++/v1/utility:251:
In file included from /usr/local/opt/llvm/bin/../include/c++/v1/compare:145:
In file included from /usr/local/opt/llvm/bin/../include/c++/v1/__compare/compare_partial_order_fallback.h:13:
In file included from /usr/local/opt/llvm/bin/../include/c++/v1/__compare/partial_order.h:14:
In file included from /usr/local/opt/llvm/bin/../include/c++/v1/__compare/weak_order.h:14:
/usr/local/opt/llvm/bin/../include/c++/v1/__compare/strong_order.h:73:44: error: expected unqualified-id
                    if (__t == 0 || _VSTD::isinf(__t)) {
                                           ^
.pio/libdeps/native/Unity/src/unity_internals.h:211:18: note: expanded from macro 'isinf'
#define isinf(n) (isnan((n) - (n)) && !isnan(n))
                 ^

////// and many more like that

If I go and remove the isnan macro from unity_internals.h. everything is starting to work again.

The upgrade in my project.

[jac4e]

I encountered the same issue. A workaround that I used was to use macros to use unity's definitions for isnan and isinf when unit testing, then use the std::isinf and std::isnan otherwise.

#ifdef UNIT_TESTING
#define _ISNAN(x) isnan(x)
#define _ISINF(x) isinf(x)
#else
#define _ISNAN(x) std::isnan(x)
#define _ISINF(x) std::isinf(x)
#endif // UNIT_TESTING

I am unsure what the implications of this workaround are, but at least it allows for the use of unity without modifying it.

[DavidSchinazi]

I just hit the same issue with Apple's default toolchain. The workaround above didn't work for me, but I was able to find another one. I added a custom unity config.h with #define UNITY_EXCLUDE_FLOAT in it. This only works for me because I don't use floating point math in unity itself, but it'll have to do for now.

[jonathangjertsen]

I encountered a similar issue when evaluating Au, which defines an isnan template function. I think this macro should be renamed to UNITY_ISNAN or similar, isnan is too generic.

Note: my workaround here was to add #undef isnan to the top of the generated au.h file. In cmake:

add_custom_target(
    au-rebuild
    COMMAND echo \#undef isnan > ${CMAKE_CURRENT_SOURCE_DIR}/prebuilt/au-prebuilt.h
    COMMAND ${PYTHON_EXE} tools/bin/make-single-file --noio --units ${UNITS} >> ${CMAKE_CURRENT_SOURCE_DIR}/prebuilt/au-prebuilt.h
    WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}/au
)

[mvandervoord]

Good point. I agree.

[mvandervoord]

Thanks everyone. I believe the cutting edge has this fixed now.

[DavidSchinazi]

Thank you! For anyone wondering, this was fixed in a1b1600e4361cdf98bdb57e7d32a1cafa6eb90bf and is included in v2.6.0 RC1.

[mvandervoord]

Thanks @DavidSchinazi ! I should have included those details. Thanks for picking up the slack!