This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [C++ Patch/RFC] Final part of c++/26099, __is_convertible_to
Paolo Carlini wrote:
> I see. My "new" plan then is trying to figure out which checks can be
> made simpler when only the types count. Hopefully I will find the code
> easy to understand from that point of view (I'm not completely
> optimistic ;) ... Besides, ambiguities, access checks, etc, should be
> pulled from somewhere, because implicit_conversion by itself doesn't do
> that.
Oh, yes, you did explain some of this to me before. The bits about
access checks and such are indeed complicated. In that case, you might
indeed want to call convert_like, but you'll want to pass down a
no-errors parameter.
C++0X, via this predicate, has introduced something new: it used to be
that we always got to decide what to do, and then checked access and
ambiguity as we did it. Now, we have to use access and ambiguity as
part of the decision process. That's inevitably going to lead to
threading an no-error parameter through all this code.
> I'm thinking that maybe we can discuss face to face some details
> of my ongoing work in Ottawa?!?
Sure, I'd be happy to chat with you.
--
Mark Mitchell
CodeSourcery
mark@codesourcery.com
(650) 331-3385 x713