allow code to build with -std=gnu23 (which is the new default in gcc-15)

[fabiangreffrath]

No description provided.

[rfomin]

-std=gnu23 (for C23 with GNU extensions).

C23 is only partially implemented in Clang: https://clang.llvm.org/c_status.html#c2x
I think C99 is just fine for us.

[fabiangreffrath]

Linux GCC and MSYS2 pass without changes. Good.

[rfomin]

Linux GCC and MSYS2 pass without changes. Good.

Should we set the standard there?

woof/CMakeLists.txt

Line 18 in 3f41e8f

set(CMAKE_C_STANDARD 99)

I think otherwise CMake will force C99 anyway.

[fabiangreffrath]

Indeed, seems that it overrides the standard at this point. I misread this as an "at least" requirement.

[fabiangreffrath]

I'll try this as an experiment.

[fabiangreffrath]

Linux GCC and MSYS2 pass without changes. Good.

Now that's the kind of trouble I was expecting. 😁

[fabiangreffrath]

Now, I consider these changes worth keeping. Is there anything we put to risk by including <stdbool.h> instead of defining our own false/true?

[rfomin]

Is there anything we put to risk by including <stdbool.h> instead of defining our own false/true?

It should work, but I prefer to avoid it. It's misleading since we have our own boolean type with int32_t size.

[fabiangreffrath]

It should work, but I prefer to avoid it. It's misleading since we have our own boolean type with int32_t size.

Right, we don't let stdbool.h define our boolean type, just false and true. We still have typedef int boolean a few lines later.

[rfomin]

Right, we don't let stdbool.h define our boolean type, just false and true. We still have typedef int boolean a few lines later.

So why do we need stdbool.h? Because of https://en.cppreference.com/w/c/language/bool_constant?

[fabiangreffrath]

Because in newer C standards, true and false are reserved key words. You can check for them being defined, though, and we do so by checking the __bool_true_false_are_defined macro. But guess what, this macro is only ever defined if you include stdbool.h and this in turn unconditionally defines the key words.

[rfomin]

LGTM.

[fabiangreffrath]

Because in newer C standards, true and false are reserved key words. You can check for them being defined, though, and we do so by checking the __bool_true_false_are_defined macro. But guess what, this macro is only ever defined if you include stdbool.h and this in turn unconditionally defines the key words.

In other words, Schroedinger's cat. You only know if the key words are defined if you make sure they are defined. 😉

[fabiangreffrath]

LGTM.

Thanks!