Re: [Boost-users] [Signals] Is connection::disconnect() "const-correct"?

Subject: Re: [Boost-users] [Signals] Is connection::disconnect() "const-correct"?
From: Frank Mori Hess (fmhess_at_[hidden])
Date: 2011-05-03 23:10:03

On Tuesday, May 03, 2011, Greg Hickman wrote:
> For a connection 'c', if c.connected() == true prior to calling
> 
>     c.disconnect();
> 
> after which c.connected() == false, how can connection::disconnect() be
> considered a "const" member function since it clearly changes the
> observable state of a connection?

Signals2 continues to do the same, although if I ever thought much about 
rationalizing it before, I don't remember doing so now.  Anyways, maybe 
it's because connection objects are just handles to the real connection.  
So it's like a pointer in that a const pointer just means the pointer can't 
be reassigned to point at another object, but you can still modify the 
pointed-at object through it.

To make things more concrete, suppose disconnect was non-const.  It still 
wouldn't stop you from making a new non-const connection object that is a 
copy of the const connection object, and then disconnecting through the 
non-const connection object.  So I guess I'm arguing that the connected() 
method isn't returning information about the state of the connection object 
itself, but rather the state of the thing it is a handle to.