Clang-Tidy complaint about re_snprintf

[juha-h]

Android Studio Clang-Tidy complains about all re_snprintf calls. Example is below. Is this a bug in Clang-Tidy or a real issue with the macro?

Clang-Tidy: Conditional operator with identical true and false expressions
Declared in: re_fmt. h
Definition:
#define re_snprintf(str, size, fmt, ...)                                      \
	_re_snprintf_s((str), (size), (fmt), RE_VA_ARGS(__VA_ARGS__))
Replacement:
_re_snprintf_s((event_buf), (sizeof event_buf), ("call closed,%s,%s"),
        _Generic((0) ? (prm) : (prm),
                _Bool: sizeof(int),
                char: sizeof(int),
                unsigned char: sizeof(unsigned int),
                short: sizeof(int),
                unsigned short: sizeof(unsigned int),
                int: sizeof(int),
                unsigned int: sizeof(unsigned int),
                long: sizeof(long),
                unsigned long: sizeof(unsigned long),
                long long: sizeof(long long),
                unsigned long long: sizeof(unsigned long long),
                float: sizeof(double),
                double: sizeof(double),
                char const *: sizeof(char const *),
                char *: sizeof(char *),
                void const *: sizeof(void const *),
                void *: sizeof(void *),
                struct pl: sizeof(struct pl),
                default: sizeof(void *)),
        (prm),
        _Generic((0) ? (tone) : (tone),
                _Bool: sizeof(int),
                char: sizeof(int),
                unsigned char: sizeof(unsigned int),
                short: sizeof(int),
                unsigned short: sizeof(unsigned int),
                int: sizeof(int),
                unsigned int: sizeof(unsigned int),
                long: sizeof(long),
                unsigned long: sizeof(unsigned long),
                long long: sizeof(long long),
                unsigned long long: sizeof(unsigned long long),
                float: sizeof(double),
                double: sizeof(double),
                char const *: sizeof(char const *),
                char *: sizeof(char *),
                void const *: sizeof(void const *),
                void *: sizeof(void *),
                struct pl: sizeof(struct pl),
                default: sizeof(void *)),
        (tone), 0)

[sreimers]

This is a false positive and a needed workaround for some compilers.

[juha-h]

So is your suggestion that I open an Android Studio ticket about this?

[sreimers]

Here are the details:
#1110
https://gcc.gnu.org/legacy-ml/gcc/2016-02/msg00266.html

But since it's a gcc bug/workaround, I'm unsure if there is a real fix for clang-tidy.

[sreimers]

Maybe a // NOLINT for clang-tidy is a option. but unsure how it have to look like and it works for this macro.

[juha-h]

I created Android Studio ticket, but usually they don't lead to anything.

[juha-h]

I got reply from Android Studio to my ticket:

I see the following in the macro definition:

_Generic((0) ? (prm) : (prm)

Which is a conditional operator, and it has prm on both sides. Clang-Tidy warning seems reasonable to me.

If I replace that line with:

_Generic((prm)

Then the Clang-Tidy warning disappears. I don't know what purpose that ternary operator serves.

So why can't the macro definition be changed like proposed?

[sreimers]

See GCC Bitfields bug: #1110

[sreimers]

You can try #1206

[juha-h]

I'm on the road. Will try when somewhere.

[juha-h]

#1206 helped and re_snprintf complaints are gone.