Talk:Persistence unified

I like the fact that object versioning is being considered, this consumes a lot of my time with current serialization methods in other languages.

One question I have though is does the actual persistance mechanism have to be done by the programmer at all? Right now there is a type of "persistance" already in the language; the compiler determines how an object should be "blobbed" in to system memory. A binary serialization could essentially be "blobbing" the exact same memory representation of the object to disk and when it is reloaded, swizzling the pointers to the new memory locations. The compiler could determine the way to "blob" the object in to an XML format breaking all objects down to types handled built in by the compiler; string versions of integers, reals, strings directly copied out, SPECIAL in a hex string format, etc.

This way when the system is compiled, a feature of ANY could be generated by the compiler itself and systems wouldn't have to be modified at all.

Is this approach insufficient? --Clemahieu 18:18, 4 October 2007 (CEST)