On Fri, Sep 1, 2017 at 2:46 PM, Marek Polacek <pola...@redhat.com> wrote:
> @@ -16960,6 +16961,24 @@ cp_parser_simple_type_specifier (cp_parser* parser,
>
> return type;
>
> + case RID_REMOVE_QUALS:
> + /* Consume the `__remove_qualifiers' token. */
> + cp_lexer_consume_token (parser->lexer);
> + /* Parse the operand to __remove_qualifiers`'. */
> + type = cp_parser_sizeof_operand (parser, RID_REMOVE_QUALS);
Wouldn't it be simpler to use cp_parser_type_id?
> + if (!TYPE_P (type))
> + {
> + error_at (token->location,
> + "%<__remove_qualifiers%> can only be applied to a type");
> + type = error_mark_node;
> + }
...rather than check this.
It doesn't look like you handle use of __remove_qualifiers in a
template, and you don't have any testcases for it.
> +@code{__remove_qualifiers} takes a typename as an argument:
I think it would be better to use the term "type-id" here, to avoid
confusion with "type-name", which is only a single identifier.
Jason
