Discussion
Weird thought.
I am building a codebase. I am getting *weird* errors. Multiple of them. Errors that don't quite make sense.
I believe I am building with C++17. Some of the errors would be less strange if I were targeting some other C++ standard.
Is there some way to get my code to "leak" the C++ standard? Something like
#error __cplusplus
That doesn't work. I *can* do this
#if __cplusplus == 201703L
#error Success
#endif
This errors success! But if it hadn't, I sure would be lost.
@mcc that does smell a bit like an MSVC issue i've encountered before, but the "attempt to create an array of negative that number" idea is excellent
@mcc could also be your system has a pre C++17 libstdc++, or at least the compiler is getting the -I flag for a "wrong" one before the one you need.
If you're not sure what the compiler arguments are, you should be able to give your build system a shell script that just dumps the args to the console.
Or just look at the compile_commands.json, if your build system can export them.
int x[-__cplusplus];
main.cpp:11:19: error: narrowing conversion of ‘-201703’ from ‘long int’ to ‘long unsigned int’ [-Wnarrowing]
11 | int x[-__cplusplus];
| ^
main.cpp:11:7: error: size ‘-201703’ of array ‘x’ is negative
11 | int x[-__cplusplus];
| ^
@mcc This may be a dumb suggestion, but wouldn't something along the lines of "cout << __cplusplus" give you what you need? Or does it specifically have to be at compile time?
@mal3aby that's a good idea, but did not work at the time I made this post as the program was at that time not compiling 😞 and we believed build system getting confused about the C++ standard level was why.
@mcc when I do this I copy a build command and run it with a preprocessor flag -E to see what the macros expanded to, so I can inspect __cplusplus in the expanded source code.
@mcc $ c++ -x c++ -dD -E - </dev/null | grep cplus
#define __cplusplus 201703L