Thinking about the Remote Object schemas

Presently (and as an interim measure, I understand) in order to communicate
with a remote object, a top level description of that object has to be
present in a config file.  Something like the following:

  {
    "id":"Remote.json",
    "sources":
    [ {
      "name":"http://server.com:8080/Toot",
      "sourceClass":"org.persvr.datasource.HttpJsonSource",
      "schema":{
        prototype:{
          remote:true,
        },
        methods:{
          methodOn:{
            runAt:"server"
          },
          methodTwo:{
            runAt:"server"
          },
        },
        properties:{
        }
      }
    ...

I'm finding this very limiting.  I now have a network of servers and it's
not convenient for each server to have a record of this kind for each other
server x class cross product.

Could I suggest that these could be generated automatically from the query?
 I.e. if I want and object at "http://server.com:8080/MyClass" the server
address is provided, it's obviously an attempt to do a HttpJsonSource, and
any methods that don't come back from the remote server are going to be
runAt:server.

If we inferred this file for now, I think Persevere would move faster into
it's rightful Web Services / Distributed Computing niche.  Later, we can
also read the remote schema as has been discussed.

We will need to work an authentication story into this sometime too.

Are there other things this file is doing / will do that I'm overlooking?

Original issue reported on code.google.com by davidthi...@gmail.com on 24 May 2009 at 5:02

kazu2012 / persevere-framework

Thinking about the Remote Object schemas #191