Re: [C++0x] avoid extra tentative parse in range-based for loops

[Paolo Bonzini <bonzini at gnu dot org>]

On 01/13/2011 12:24 PM, Rodrigo Rivas wrote:
> On Thu, Jan 13, 2011 at 10:04 AM, Paolo Bonzini<bonzini@gnu.org>  wrote:
> > > > Then, what about adding somewhere?
> > > > TREE_USED (range_decl) = 1;
> > > > DECL_READ_P (range_decl) = 1;
> >
> > Replying to an old message: does __attribute__ ((__unused__)) work in this
> > context?  Or also, "for (Work : { w(), w(), w() })" could be accepted as a
> > GCC extension.
>
> Actually, it does work! If I remove these two lines, then the code:
>        for (int a __attribute__((unused)): { 1, 2, 3 }) { }
> gives no warning at all with -Wall.

Yeah, it's just a declaration.  I was only concerned about the 
subtleties of the parser and whether it accepted attributes at all.

> So, maybe that two lines are unnecessary after all.

Yes.

> And the same goes for the nameless-variable extension.

It could still be useful to propose it to the committee, though Jason 
said it is late for C++0x.

Paolo