[Bug tree-optimization/87034] [9 Regression] missing -Wformat-overflow on a sprintf %s with a wide string

[msebor at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87034

--- Comment #6 from Martin Sebor <msebor at gcc dot gnu.org> ---
The warning in GCC 8 isn't bogus, it just reports the wrong number of bytes (it
counts characters, not bytes).  That the warning is no longer issued is a
regression caused by c_strlen() giving up when the size of the constant array
element doesn't match the expected size (in this case, sizeof(wchat_t) vs
sizeof(char)).  That change was made out of an abundance of caution in an
effort to avoid folding suspect expressions that were felt didn't need to be
folded (e.g., strlen((char*)L"\x41424344\x45464748")).  There's nothing
inherently wrong with folding those expressions, but they don't need to be
folded for GCC to understand their length and make use of it to issue warnings.