Parser: add warning for using UTF-8 char literal before C23

src/Diagnostics.zig

@@ -2568,6 +2568,12 @@ const messages = struct {
         const kind = .warning;
         const suppress_version = .c23;
     };
+    pub const u8_char_lit = struct {
+        const msg = "UTF-8 character literal is a C23 extension";
+        const opt = "c23-extensions";
+        const kind = .warning;
+        const suppress_version = .c23;
+    };
 };
 
 list: std.ArrayListUnmanaged(Message) = .{},

src/Parser.zig

@@ -7684,6 +7684,7 @@ fn charLiteral(p: *Parser) Error!Result {
             .node = try p.addNode(.{ .tag = .char_literal, .ty = Type.int, .data = undefined }),
         };
     };
+    if (char_kind == .utf_8) try p.err(.u8_char_lit);
     var val: u32 = 0;
 
     const slice = char_kind.contentSlice(p.tokSlice(p.tok_i));

test/cases/u8 character constant.c

@@ -1,6 +1,6 @@
 const unsigned char c = u8'A';
 _Static_assert(c == 'A', "");
-
+#pragma GCC diagnostic ignored "-Wc23-extensions"
 _Static_assert(u8'\xff' == 0xFF, "");
 
 const unsigned char c2 = u8'™';
@@ -12,7 +12,9 @@ const unsigned char c4 = u8'AA';
     #error Character constant should be true in preprocessor
 #endif
 
-#define EXPECTED_ERRORS "u8 character constant.c:6:26: error: character too large for enclosing character literal type" \
+#define EXPECTED_ERRORS \
+    "u8 character constant.c:1:25: warning: UTF-8 character literal is a C23 extension [-Wc23-extensions]" \
+    "u8 character constant.c:6:26: error: character too large for enclosing character literal type" \
     "u8 character constant.c:7:26: error: character too large for enclosing character literal type" \
     "u8 character constant.c:8:26: error: Unicode character literals may not contain multiple characters" \
 

test/cases/wide character constants.c

@@ -1,4 +1,4 @@
-//aro-args -Wfour-char-constants
+//aro-args -Wfour-char-constants -Wno-c23-extensions
 /*
 
     A multiline comment to test that the linenumber is correct.