Lies in education [was Re: The "loop and a half"]

[Marko Rauhamaa]

Chris Angelico <rosuav at gmail.com>:

> The places where C++ is not a superset of C are mostly things you
> wouldn't want to be doing anyway. You can generally take C code and
> compile it with a C++ compiler, and it'll have the same semantics.

Here's a C/C++ program:

========================================================================
#include <stdio.h>

int main()
{
    struct {} s;
    printf("%d\n", (int) sizeof 'a');
    printf("%d\n", (int) sizeof s);
    return 0;
}
========================================================================

When compiled (with gcc) as a C program, the output is:

    4
    0

When the same program is compiled (with gcc) as a C++ program, the
output is:

    1
    1

That is not immediately all that significant but points to subtle
incompatibilities between the data models of C and C++.

Then we have syntactic problems:

========================================================================
int main()
{
    void *s = "hello";
    char *t = s;
    return 0;
}
========================================================================

which, as a C program, makes gcc perfectly happy, but a C++ compilation
complains:

   test.C: In function ‘int main()’:
   test.C:5:15: error: invalid conversion from ‘const void*’ to ‘void*’ \
   [-fpermissive]
        void *s = "hello";
                  ^~~~~~~
   test.C:6:15: error: invalid conversion from ‘void*’ to ‘char*’ [-fper\
   missive]
        char *t = s;
                  ^

(The first one should trigger an error message even in C compilation,
but string literals were standardized to be "semiconstant" in C.)


Marko