Re: [sigc] Proposal for standardization in C++ Library TR2

[Carl Nygard <cjnygard verizon net>]

On Fri, 2005-07-29 at 13:41 -0500, Doug Gregor wrote:
> On Jul 29, 2005, at 11:06 AM, Lars Gullik Bj� wrote:
> > I think there is more than two names that needs to be decided on.
> >
> > With signal/slot we have at least: signal, slot, connection and emit
> > (wheter the explit emit will be in the standard or not is a different
> > subject)
> >
> > For publisher/subscriber this could be: publisher, subscriber,
> > subscription and publish
> >
> > Are there other concepts in this framework that needs to be named?
> 
> There's the "trackable" class, which could potentially get a different 
> name, but it doesn't have to. Signal, slot, and connection are the big 
> ones.
> 
> Murray, Carl: How do you feel about this terminology?

I've made my thoughts known on this.  The only problem I have with
publisher/subscriber is that the class (formerly known as Slot)
std::subscriber isn't *really* the subscriber, it's just a link to the
actual subscriber.  Hence the naming of std::proxy_fun as the Slot
object.

However, if no one has jumped on that bandwagon, I'll not stand in the
way of consensus.

So we'd have:

std::publisher
std::subscriber
std::connection
std::trackable

I'd be curious to see how close libsigc++/Boost.Signals is to the
idealized GoF Publisher/Subscriber pattern (not that I want to induce
feature creep).  I thought I had the book here in my library, but I
guess it's at work.  Oh well, excuse to go to Borders and grab a
cup-o-joe.

Regards,
Carl