test(ezc): comprehensive test coverage

ezc/Makefile

@@ -47,7 +47,7 @@ $(BIN): $(OBJ)
 	$(CC) $(CFLAGS) $(INCLUDES) -c -o $@ $<
 
 clean:
-	rm -f $(OBJ) $(BIN) tests/test_lexer tests/test_parser
+	rm -f $(OBJ) $(BIN) tests/test_lexer tests/test_parser tests/test_typechecker tests/test_codegen
 
 install: build
 	install -d /usr/local/bin
@@ -73,14 +73,27 @@ tests/test_lexer: tests/test_lexer.c $(TEST_DEPS)
 tests/test_parser: tests/test_parser.c $(TEST_DEPS)
 	$(CC) $(CFLAGS) $(INCLUDES) -o $@ $^ $(LDFLAGS)
 
-test-unit: $(TEST_DEPS) tests/test_lexer tests/test_parser
+tests/test_typechecker: tests/test_typechecker.c $(TEST_DEPS)
+	$(CC) $(CFLAGS) $(INCLUDES) -o $@ $^ $(LDFLAGS)
+
+tests/test_codegen: tests/test_codegen.c
+	$(CC) $(CFLAGS) $(INCLUDES) -o $@ $< $(LDFLAGS)
+
+test-unit: $(TEST_DEPS) tests/test_lexer tests/test_parser tests/test_typechecker
 	@echo ""
 	@echo "=== Lexer Tests ==="
 	@./tests/test_lexer
 	@echo "=== Parser Tests ==="
 	@./tests/test_parser
+	@echo "=== Type Checker Tests ==="
+	@./tests/test_typechecker
+
+test-e2e: build tests/test_codegen
+	@echo ""
+	@echo "=== End-to-End Codegen Tests ==="
+	@./tests/test_codegen
 
 test-parity: build
 	@./tests/run_parity.sh
 
-test: test-unit test-parity
+test: test-unit test-e2e test-parity

ezc/src/parser/parser.c

@@ -683,15 +683,34 @@ static AstNode *parse_func_declaration(Parser *p) {
              *   -> (int, string)        plain types
              *   -> (x int, y int)       named returns
              *   -> (x, y int)           shared type
+             *
+             * Disambiguation: if the identifier is a known type name,
+             * it's a plain type list, not names.
              */
             next_token(p);
             while (!cur_token_is(p, TOK_RPAREN) && !cur_token_is(p, TOK_EOF)) {
-                if (cur_token_is(p, TOK_IDENT) && peek_token_is(p, TOK_IDENT)) {
+                /* Check if current ident is a type name (not a variable name) */
+                bool is_type = false;
+                if (cur_token_is(p, TOK_IDENT)) {
+                    const char *lit = p->cur_token.literal;
+                    is_type = (strcmp(lit, "int") == 0 || strcmp(lit, "uint") == 0 ||
+                        strcmp(lit, "i8") == 0 || strcmp(lit, "i16") == 0 ||
+                        strcmp(lit, "i32") == 0 || strcmp(lit, "i64") == 0 ||
+                        strcmp(lit, "u8") == 0 || strcmp(lit, "u16") == 0 ||
+                        strcmp(lit, "u32") == 0 || strcmp(lit, "u64") == 0 ||
+                        strcmp(lit, "float") == 0 || strcmp(lit, "f32") == 0 ||
+                        strcmp(lit, "f64") == 0 || strcmp(lit, "string") == 0 ||
+                        strcmp(lit, "bool") == 0 || strcmp(lit, "char") == 0 ||
+                        strcmp(lit, "byte") == 0 ||
+                        (lit[0] >= 'A' && lit[0] <= 'Z')); /* struct/enum types */
+                }
+
+                if (cur_token_is(p, TOK_IDENT) && peek_token_is(p, TOK_IDENT) && !is_type) {
                     /* Named return: name type — skip name, use type */
                     next_token(p);
                     node->data.func_decl.return_types[node->data.func_decl.return_type_count++] =
                         p->cur_token.literal;
-                } else if (cur_token_is(p, TOK_IDENT) && peek_token_is(p, TOK_COMMA)) {
+                } else if (cur_token_is(p, TOK_IDENT) && peek_token_is(p, TOK_COMMA) && !is_type) {
                     /* Shared type: (x, y int) — count names, assign same type to all */
                     int shared = 1;
                     while (peek_token_is(p, TOK_COMMA)) {