From: Daniel Frey (d.frey_at_[hidden])
Date: 2002-07-08 13:07:33
On Mon, 2002-07-08 at 11:56, David Abrahams wrote:
>
> > For the doc, there is a simple problem: The docs say that the Supplied
> > Operations looks like this:
> >
> > T operator+(T, const T&)
> >
> > changing the return type to const is straight forward, but the
> > parameters depend on the compiler (or even worse, the compiler version
> > and the user-config). I currently plan to replace the above line with
> >
> > const T operator+(const T&, const T&)
> >
> > and add a footnote to all operators that depend on the NRVO-settings.
>
> OK. In the standard we could handle that with the as-if rule and the fact
> that the user can't reliably take the address of a library-supplied
> function, but your way seems simpler.
Well, I added the option to force the implementation to create the
documented signature, you just have to set
BOOST_FORCE_SYMMETRIC_OPERATORS. It would IMO be an unnecessary
restriction to forbid taking the address.
Currently, I have a 36K patch that should be appliable to the
boost_1_28_0 directory. It adds the new operator.hpp, regression tests
and documentation. It only patches the compiler-config for the gcc, as
it is the only compiler that has the NRVO. I use a positive indicator
(BOOST_HAS_NRVO) instead of a negative (BOOST_NO_NRVO) for simplicity
now. Shall I post the patch here, send it to you privately or check it
in to the files section?
Regards, Daniel