Re: Deprecate g++ overloading extension?

[Mark Mitchell <mark at codesourcery dot com>]

Jason Merrill wrote:

> On Fri, 2 Jul 2004 10:13:58 -0700, Mark Mitchell <mark@codesourcery.com> wrote:
>
>  
>
> > We have an undocumented extension to overload resolution in "joust".
> > Unless we're being pedantic, we make a try at resolving ambiguities by
> > choosing the function whose worst conversion is least bad.
> >
> > I haven't seen any sign that the committee is considering extending
> > the overload rules in this way, and since this is totally
> > undocumented, people using this alleged feature are presumably doing
> > so purely by accident.
> >
> > What do you think about removing this extension from non-pedantic
> > compilation?
> >    
>
> I'm opposed to removing it.  We already give an unconditional pedwarn which
> tells the user what's wrong with their code.
Actually, that's not true.

If we pick a built-in candidate, we don't warn -- because we never go 
through the call machinery which is where the warning would be generated.

That's why this came up; I talked to someone who was confused by the 
fact that G++ did not match another compiler's behavior.

Will you agree to implement a fix for the case where we pick a built-in 
candidate?

--
Mark Mitchell
CodeSourcery, LLC
(916) 791-8304
mark@codesourcery.com