The previous post, Business Logic Customization and the Four Plugin Stages , closed by saying the level changes here. From this point the subject is no longer what a definition declares, it is how those definitions and that data travel once they leave memory.
Serialization is two separate jobs. One is persistence: an object has to be stored and read back unchanged next time, which is how definitions are stored and how system data such as SessionUser is stored. The other is transport: an object has to reach the other end, and the other end is not one thing. Desktop, browser, iOS and Android all run in different environments.
This post covers:
Which route persistence takes, why transport splits into two, and what the dividing line is drawn on
How PayloadFormat decides what a call's value looks like on the way out
Why definition style things go out as an XML string rather than a JSON tree
Why not one transport type can be left unregistered
On the deserialization end, who decides the type, and why onlyβ¦