Martin Slater wrote:
Robert Ramey wrote:
I can't see why this would require a new archive type
Now that would be perfect if I did not need a new archive type, I'm having trouble envisioning how it would work without one though. Any chance you could explain a bit more?
thanks
Martin
I presumed the following: a) That "document" is represented by a collection of other objects b) and that commands act upon this collection by 1) inserting elements 2) deleting elements 3 changing the properties of elements. c) and that the elements have some sort of commonality. They might all be one of the following 1) descendants of a common base class 2) instances of boost::variant (serialization now in main cvs) 3) instances of boost::any (serialization pending) Assuming this is the case, the serialization of this "document" is independent of any archive implementation. and serialization is straightforward. Indeed, if the collection is one of the STL collections, then its already done. Robert Ramey