Someone should tell the gcc maintainers then.

    int main()
    {
        int a = 12;
    }

    $ gcc maintest.c
    $ ./a.out
    $ echo $?
    148

  $ cat test.c
  int main(void) {int a = 12;}
  $ gcc-6 test.c -o test --std=c89; ./test; echo $?
  162
  $ gcc-6 test.c -o test --std=c99; ./test; echo $?
  0

Must be a C99 thing.

This isn't that conclusive of a test -- uninitialized memory is 0 relatively often.

Okay, that is conclusive. (And I'm pleasantly surprised by how readable GCC's source code is.)

This has nothing to do with memory. The x86_64 ABI returns values from functions in the RAX register.

Sure; same point applies though -- just because 0 happens to be in RAX doesn't mean it's defined behavior.

Just tested on the MSVC compiler I have at hands. The standard is not respected.

MSVC doesn't claim to implement C99, where this rule was added, so it kind of makes sense that it doesn't happen. (generally, if they added C99-stuff then only where it was required for C++)

    int main(void)
    {
        __asm {
          mov eax, 123
        }
    }

    // visual studio command line
    // cl testc.c
    // cl testcpp.cpp
    // testc.exe
    // echo C %errorlevel%
    // testcpp.exe
    // echo CPP %errorlevel%

Indeed. It's a C99 thing.

For those not familiar with MSVC. The C compiler is not C99 compliant and the C++ compiler is.

I love C but this whole thread highlights a lot of the criticism of it. Everyone's right based on some standard or switch, and everyone's wrong for the same reason.

This is not specific to C, but common to all evolving standards. If you want portable code just drive on the middle of the road.

Often this is as easy as coding to an older standard, like C89 with a few selected features from later standards, and adding some compatibility features / problem detection in the build system.

And don't rely on features that are difficult to explain and/or add only questionable value. Just return 0 from main, it's the obvious simple thing to do.