Thank you for your input everyone.
@sugoidev , my initial thought was that the TextAsset act only as a reference and be a copy/mirror of the original serialization (stored in the graph), so the canonical serialization always be the graph serialization instead of the Text Asset. This will make things easier (as in less prone to bugs and data loss in case TextAsset gets deleted for whatever reason), as well as avoid extra options in inspectors (which I am never fond of :P).
Do you guys thnk that having a “Serialization Source” option (choice between internal json or external TextAsset) would be important and helpful?
Gavalakis
