RFC: Deprecate (a little bit) of C++ attribute syntax

[Gabriel Dos Reis]

Mark Mitchell <mark@codesourcery.com> writes:

| Currently, the C++ front end accepts this syntax:
| 
|   int x (3) __attribute__((unused));

[...]

| Therefore, I'm proposing deprecating this syntax in GCC 3.3 and
| removing it in GCC 3.4 (with the new parser). 

|  I'm not sure if I can
| get the old parser to actually warn about this (it's confused and
| thinks the parenthesized initializer is part of the declarator, but
| there is then later magic to untangle that!)

Do you think it is the same reason why it seems to ignore
attributes for nested typedef-names? I.e. in

   struct A {
      struct X { } __attribute__((__aligned__(4)));
      typedef struct { } Y __attribute__((__aligned__(4)));
   };

__alignof__(A::X) == 4 but __alignof__(A::Y) == 1.
?

| Objections?

I mostly agree with all of yours suggestions -- as you know, I've been
working on the attribute stuff these days.  The oddity you mention may
be more prenicious.  Consider:

   template<typename T>
      struct X { } __attribute__((__aligned__(__alignof__(T))));

This may, at first sight, look innocent, but it doesn't do the obvious
thing one thinks of.

The problem here is that at the closing brace, the template-parameter
"T" goes out of scope -- the current parser doesn't diagnose it
because we don't do 2-phase name lookup yet.

-- Gaby