No, not really. In the API, addresses are always part of some other entity (an order’s delivery address for example).
The geocoding API is there to validate user typed addresses and to find lat/lng coordinates to be copied over to an entity. (similar to google’s geocoding service)
In what way is the client misled? Because the client thinks it knows an ID that the API / server uses? But the client cannot use this ID (it can’t lookup an address using the ID). Although it’s a leaky abstraction I’m struggling to see what potential problems this could lead to.
Exactly. I think you are answering your own question It can be done for sure. The server can generate transient and unique ID numbers on the fly and maybe there are no real problems with this. But as you say, it’s a leaky abstraction and misleading the client to think that these IDs have some meaning and can be referenced to later on.
I guess I was just hoping that I missed something in the spec and that there actually was a nicer way of doing this that I just hadn’t realized.