------- Comment #2 from jakub at gcc dot gnu dot org 2008-10-31 10:59 -------
Certainly e.g. cp_parser_cast_expression and cp_parser_postfix_expression
could reject types with type_uses_auto, but doesn't [dcl.spec.auto] specify
only very limited number of places where auto can be used and the rest:
"A program that uses auto in a context not explicitly allowed in this section
is ill-formed."?
Enumerating all the places which parse types and auto isn't explicitly allowed
would be IMHO difficult, wouldn't it be better to add a new flag (say to
enum cp_parser_flags, CP_PARSER_FLAGS_AUTO_ALLOWED) and pass it down to
cp_parser_type_specifier*? Guess we might still need a type_uses_auto check
in a couple of places, but it will be certainly less than if we parse auto all
the time and let all the callers check for it.
E.g. ATM g++ parses without any errors:
#include <typeinfo>
const std::type_info &foo ()
{
return typeid (auto);
}
const std::type_info &bar ()
{
return typeid (auto *);
}
and returns _ZTIT_ and _ZTIPT_, which looks wrong.
BTW, seems a lot of cp_parser_type_specifier* callers still think the flags
argument is boolean and pass /*is_condition=*/false or /*is_condition=*/true to
it.
--
jakub at gcc dot gnu dot org changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |dodji at gcc dot gnu dot org
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37962
