So this just came to my mind when I was thinking about data serialization and its resemblance to windows hibernation. When you hibernate the system, the OS does not care about individual programs and whether they can serialize/deserialize their data. It just dumps the whole thing to disk and later on you can resume whatever you have been doing.
Here’s the question: How does Windows do this without caring about each individual program? Is it possible to somehow emulate this for your code so that you can “dump” it to disk and later on resume it without bothering to write serialization/deserialization methods?
Windows does this by suspending execution of every process and writing out the active (allocated) memory pages to disk. When this memory is later restored and the kernel kick-started, it is able to resume everything where it left off, because from its perspective, the memory hasn’t actually changed. It’s as though it just froze for a very long period of time.
The only way you could do this with one of your own processes would be to have some other supervisory code running in the kernel — you’d need a way to get at your process’ memory map and preserve it along with the actual memory pages so that all existing pointers in the application’s memory remain valid when the pages are restored later. You would also need a way to persist other data (such as any open file descriptors) so that they could be restored as well.
This is not practical for most applications.