Talk:Object serialization

Serializing object instances which have no state does not seem very meaningful. You could trivially satisfy this task in some languages merely by naming the objects.

Meanwhile, when I think about this issue, a variety of possibilities occur to me, for state. We could even introduce stateful classes and then [by implication] the classes would also need to be serialized -- but I do not think most of the examples do anything like this.

Can we update this task with a little bit of state? Perhaps, we can add a method which reports the time the object was first created? Or would that break all the existing implementations and thus be a bad idea?

Rdm 20:25, 2 September 2009 (UTC)

Try serializing an object graph; the inter-object links would be a reasonable piece of state. Or you could use the labels on the nodes as the state. I'd keep the construction short though; this is a serialization task, not a "build a graph" task after all. —Donal Fellows 05:38, 3 September 2009 (UTC)

For non-object oriented languages, could we allow an object alternative? The serialization of something of equal complexity to an object, like a closure. I understand it is possible to create a new task, but I would like to avoid the linked list situation. User:bengt Mon Oct 21 21:37:29 CEST 2013

If you still feel like doing this, note that even languages which are not themselves "object oriented" can implement something that could be called "objects" and "classes". For example, a C module can be treated as an object with relatively minimal effort (but serializing such a thing would require additional effort of some sort, and might also depend on platform-specific details).

Since Erlang has serialization built in, this could be a pretty example. If I have to add objects and classes (not built in) it would hide the interesting bits. The Algol example seems to use records, not objects. If my understanding is correct, than I could base Erlang on that. User:bengt Tue Oct 22 09:45:55 CEST 2013