=?UTF-8?B?Ikx1w61z?= Marques" <luis luismarques.eu> writes:

Is there any particular reason why this is accepted? (I
introduced it by mistake):

      void foo(int = 3) {}

I guess it could be useful to ensure binary compatibility when
you expect to add the parameter later?

"Steven Schveighoffer" <schveiguy yahoo.com> writes:

On Tue, 17 Jun 2014 11:15:43 -0400, Lu=C3=ADs Marques <luis luismarques.=
eu>  =

wrote:

 Is there any particular reason why this is accepted? (I
 introduced it by mistake):

       void foo(int =3D 3) {}

 I guess it could be useful to ensure binary compatibility when
 you expect to add the parameter later?

Of course this should be accepted. Omitting the name has nothing to do  =

with the API, but has to do with the fact that the implementation doesn'=
t  =

use it. An implementation may be constrained by a base class or by some =
 =

duck-type requirement even though it doesn't use the parameter.

-Steve

"Steven Schveighoffer" <schveiguy yahoo.com> writes:

On Tue, 17 Jun 2014 11:19:46 -0400, Steven Schveighoffer  
<schveiguy yahoo.com> wrote:

 Of course this should be accepted.

I want to apologize, that came out sounding condescending, I didn't mean  
it that way.

-Steve

=?UTF-8?B?Ikx1w61z?= Marques" <luis luismarques.eu> writes:

On Tuesday, 17 June 2014 at 15:25:22 UTC, Steven Schveighoffer
wrote:
 Of course this should be accepted.

 I want to apologize, that came out sounding condescending, I 
 didn't mean it that way.

No problem. Thanks for clarifying this for me, I was just wanted
to understand if/why this was useful (most likely scenario), or
if it was similar to those cases where the frontend accepts
nonsensical declaration qualifiers.

"Maxim Fomin" <maxim maxim-fomin.ru> writes:

On Tuesday, 17 June 2014 at 15:15:44 UTC, Luís Marques wrote:
 Is there any particular reason why this is accepted? (I
 introduced it by mistake):

      void foo(int = 3) {}

 I guess it could be useful to ensure binary compatibility when
 you expect to add the parameter later?

Actually there is nothing strange because current implementation 
technically does not remove variable names, it generates implicit 
ones. This compiles:

void foo(int = 0)
{
	_param_0 = 1;
}

and is equivalent to

void foo(int _param_0 = 0)
{
	_param_0 = 1;
}

It is not a wise design decision which is aimed to support some 
case, it is just technical consequence of implementation.

And I don't think that it has anything to do with binary 
compatibility, because both parameter names and default arguments 
exists only in compile time.