Re: any_function (callback) library

From: williamkempf_at_[hidden]
Date: 2001-04-10 08:45:16

--- In boost_at_y..., jk_at_s... wrote:
> 9 Apr 2001 19:16:12 +0400 Kostya Altukhov wrote:
> >I think in C++ callbacks to member functions are far more important
> >than callbacks to free functions. I find it a little bit strange 
for C++
> >callback library not to support callbacks to member functions.
> 
> It is easy to produce fuction object from member function with 
binders.
> 
> >Talking about callbacks, what I really would like to be a part of 
boost
> >is libsigc++. Unfortunately it does not seem to be possible because
> 
> I don't think that libsigc++ is very good, unless I'm missing 
something,
> because it allows to call back into freestanding function only, not 
into
> arbitrary function object. It is possible to register callback into 
member
> function with the pointer to object, but this possibility is 
intrusive - it
> requires to derive object from SigC::Object. It is not good IMO.

I'd suggest taking another look at libsigc++.  It doesn't really have 
the problems you're alluding to here.  In any event, what's wanted is 
a signals/slots mechanism for Boost which is a good idea even if 
libsigc++ had the faults you suggest (i.e. the concept would still be 
worth while, though a change to the implementation would be warranted 
if you were correct about libsigc++ constraints on use).  I think 
that Boost will eventually have such a signals/slots library, 
possibly built on any_function, or possibly a ground up 
implementation.  Check out the list archives for the lengthy 
discussion we had on this topic several months ago.

Bill Kempf