https://gcc.gnu.org/bugzilla/show_bug.cgi?id=85602
--- Comment #3 from Martin Sebor <msebor at gcc dot gnu.org> --- I agree that's not an improvement. Would something like this be better? (at least until utmp_ent is marked nonstring and GCC taught to suppress the diagnostic) size_t utmpsize = sizeof UT_ID (utmp_ent); char *comment = xmalloc (strlen (_("id=")) + utmpsize + 1); strcpy (comment, _("id=")); strncat (comment, UT_ID (utmp_ent), utmpsize); I'll try to remember to test coreutils with new warnings in the future.