https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62181
--- Comment #17 from Eric Gallager <egallager at gcc dot gnu.org> --- (In reply to Xi Ruoyao from comment #16) > (In reply to Jonathan Wakely from comment #15) > > Was this question ever answered? > > https://gcc.gnu.org/ml/gcc-patches/2017-06/msg01337.html > > Oh that's intentional. > > This would make this warning more useful, while most people won't use a > "char" to address an array or something like. For example, if someone has > wrote a string-like class but forgot to overload operator+=, this warning > will detect it when he writes "char c = getchar(); buggy_string t = s + c;". > > And for something like > > "char *p = s; char x; scanf("%hhd", &x); p = p + x;" > > 1. In C++ __INT8_TYPE__ is not __CHAR_TYPE__ so there will be no warning. > He can use "int8_t x;" instead of "char x;". > 2. He can use p = &p[x], which is more clear and nobody will think this is > an append. > > I remember I wrote a response, with an option to split this into > -Wstring-plus-int=1 and -Wstring-plus-int=2. But why it wasn't sent? I > can't remember. It was in other branches of the thread: https://gcc.gnu.org/ml/gcc-patches/2017-06/msg01345.html https://gcc.gnu.org/ml/gcc-patches/2017-06/msg01347.html Again, as I always do when people propose new numeric levels to warning flags, I'd prefer splitting into new named options instead of adding numeric warning levels, for individual controllability, and end-user clarity. So in this case, maybe the final set could be: -Wstring-plus-int -Wstring-plus-char (-Wstring-plus-any-char?) -Wstring-plus-char-literal
