On 10/21/2010 12:13 PM, Joel de Guzman wrote: > > Do you mean that it has to be exactly that to be called a "visitor"? > How exact? Does it have to be "an OO hierarchy with a virtual "Dispatch" > member that accepts a visitor"?
No, there's some wiggle room. I tried to state in a latter message what I think the essence of the visitor design pattern is. > Does the Boost Variant visitation scheme (for example) satisfy these > conditions? The essentials are there, yeah. There is an implicit hierarchy, the root of which is, say, variant<A,B,C>. The subtypes are A, B and C. It's not a classic OO hierarchy, but admittedly that's not essential. There is a visitor. There are two dispatches: one on the "dynamic" type of the variant, either A, B, or C. There is second dispatch based on the type of the visitor. Pretty straightforward. Can you do a similar analysis for Thomas' code? I tried and am not certain about my results. -- Eric Niebler BoostPro Computing http://www.boostpro.com _______________________________________________ proto mailing list proto@lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/proto