https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121563
--- Comment #7 from Christopher Bazley <Chris.Bazley at arm dot com> --- It looks as though parameter forward declarations aren't marked as such until a ; is encountered by the parser, therefore a parameter forward declaration is treated like an ordinary parameter declaration for the purpose of detecting redeclaration of parameters. (push_parm_decl is called before mark_forward_parm_decls in c_parser_parms_list_declarator.) This is what I expected when I wrote https://www.open-std.org/jtc1/sc22/wg14/www/docs/n3433.pdf and it is one of several reasons I had for preferring a single semicolon-terminated list of forward parameter declarations: forward parameter declarations are still parameter declarations and not block-scope declarations, therefore they shouldn't have the same syntax as block-scope declarations (otherwise void f2a(int x, x; x) really would be valid). Consequently, the syntax proposed by N3433 implies that a forward declaration list is parsed like an ordinary parameter declaration list. The prevailing sentiment in WG14 at the moment seems to be that code which looks as though it has a bug in it (e.g. useless and has no conceivable reason to have been written like that) should be diagnosed. Given that forward declaring the same parameter twice is useless, diagnosing it seems desirable. On the other hand, I'm finding it difficult to imagine circumstances in which forward-declaring one parameter twice instead of forward-declaring two parameters would not cause other diagnostic messages. I think you'd need at least two bugs, e.g. int yin; void f(int xin; int xin; // should be diagnosed but isn't int x[xin], int y[yin], int xin, int yin);
