Re: [boost] namespace organization (was Poisoning Koenig lookup)

From: David Abrahams (dave_at_[hidden])
Date: 2002-10-29 23:19:07

"George A. Heintzelman" <georgeh_at_[hidden]> writes:

Herb Sutter is currently suggesting some guidelines for partitioning
namespaces which are very similar to George's idea. I'll let Herb give
the details, though. In general, given the current state of affairs in
the language I support this _kind_ of approach (though I haven't looked
too closely at George's proposal).

I agree with Peter that it doesn't help you to be able to use short
names safely, but I do think it can help reduce the chances of
unintended ADL.

> Peter Dimov wrote:
> > > I understood that the idea of Koenig lookup
> > > was to allow me to use those kinds of short names without having to say
> > > myprefix_shortname(iterator) every time.
> > 
> > No, the idea of Koenig lookup is to let unqualified calls work. It doesn't
> > allow shorter names.
> 
> But one of the points of namespaces is so that I can use shorter 
> (though still descriptive, in context) names without causing clashes. 
> Koenig lookup was to help make namespaces work better to encapsulate 
> interfaces. Seems to me they ought to work together. If we can't make 
> it work, okay, but it seems to me we can.
> 
> > When you introduce a function 'type' that is intended to take advantage of
> > ADL, you are effectively claiming ownership on the name 'type' in all
> > reachable namespaces. 
> 
> Yes, this is so. The question is, what should the set of reachable 
> namespaces be? Do we really want deriving from boost::noncopyable to 
> pull in the whole of boost's interface? That's really what I'm arguing 
> against in the end.
> 
> > You could argue that those reachable namespaces should
> > change to not include 'type', but the typical approach is to fix the problem
> > on your end.
> 
> So you're suggesting that because someone derives from one of these 
> boost utility classes, we've now effectively put the entire boost 
> namespace off-limits for use by arg-dependent lookup in the boost 
> user's namespaces? Because if I'm claiming ownership of the name for 
> that use in my namespace, so equally is boost implicitly claiming 
> ownership of that name for arg-dependent lookup -- its use, even though 
> not intended to be arg-dependent, prevents anyone else's.
> 
> Since these boost classes are _intended_ to be derived from, this seems 
> wrong to me. And it's not that hard to backwards-compatibly fix on 
> boost's end, by moving the base classes down into included 
> sub-namespaces which encapsulate those classes' intended interfaces. 
> Moreso since library code IMHO should be more careful about 
> unnecessarily imposing restrictions on user code.
> 
> George Heintzelman
> georgeh_at_[hidden]
> 
> 
> _______________________________________________
> Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
> 

-- 
                    David Abrahams
dave_at_[hidden] * http://www.boost-consulting.com