Richard L. Aurbach wrote:
Don,
I can't speak to what support is present in boost serialization, but I have a lot of code that needs to deal with multiple versions of classes or data structures. I've found that the best way to deal with that is to store a version number as a data member and then handle the schema evolution internally in the class. In general, it is the responsibility of a class to handle older versions "appropriately" whatever that means to you in your context. If the version number is the FIRST data member of the class (in an I/O or serialization context), then you can easily write code to handle multiple versions during I/O and/or serialization.
Just my $0.02.
This is exactly the way it's handled by the serialization library. The only difference is that the version # is included by default rather than the programmer having to explicitly do it. Robert Ramey