UI Communicating with the Host


Conversely, the UI creates a proxy much like the host does.

Remember there are two processes running, one local and one remote. The remote in the case of the UI is the host. Now let's have a look at what happens on the local side during the processing of a single plug-in from the user interface.

  1. User hits publish
  2. UI elements are parsed into proxies
  3. Proxies are sent via pyblish-rpc.client from the GUI
  4. Proxies are recieved via pyblish-rpc.service on the remote
  5. Proxies are converted into objects and passed to process() for processing.


In Maya, once pyblish_maya.setup() has been run, this is how the GUI connects with it.

import xmlrpclib
proxy = xmlrpclib.ServerProxy("")
# {'data': {'cwd': 'C:\\Program Files\\Autodesk\\Maya2015', 'pyblishRPCVersion': '0.2.0', 'currentFile': '.', 'results': 'Not supported', 'connectTime': '2016-03-03T06:32:26.449000Z', 'pythonVersion': '2.7.3 (default, May  8 2013, 09:43:48) [MSC v.1700 64 bit (AMD64)]', 'date': '2016-03-03T06:27:08.405000Z', 'host': 'maya, mayapy, mayabatch, python', 'pyblishServerVersion': '1.3.1', 'user': 'marcus', 'workspaceDir': 'C:\\Users\\marcus\\Documents\\maya\\projects\\default', 'current_file': '.', 'workspace_dir': 'C:\\Users\\marcus\\Documents\\maya\\projects\\default', 'port': 9001}, 'children': []}

Other integrations exhibit similar interfaces.

Here are some of the functionality exposed via the proxy.

Command Description
test() Return currently registered test
ping() Used to check connectivity
stats() Return statistics about the API
reset() Run collection
context(int) Return current context
discover() Return discovered plug-ins
emit() Remotely emit an event


Because Python in one process does not have access to the local variables of another process, the delegation involves a conversion of the local variables and instantiated classes into something we can transfer without loss of information and also use in order to recreate the exact scenario on the remote; JSON.

The conversion looks something like this.

class format_instance(instance):
  """Capture the essence of an Instance, just enough to visualise in a UI"""
  return {
      "name": instance.name,
      "id": instance.id,
      "children": children,
      "data": format_data(instance.data)

This is then the data used in the UI, in place of the original instance on the remote. Locally, the data is wrapped up in a class called InstanceProxy which looks something like this.

class InstanceProxy(pyblish.api.Instance):
    def from_json(cls, instance):
        self = cls(instance.pop("name"])
        copy = instance
        copy["data"] = copy.pop("data")
        self[:] = instance["children"]
        return self

    def to_json(self):
        return {
            "name": self.name,
            "id": self.id,
            "data": self.data,
            "children": list(self),
Mon Jul 18 2016 19:42:18 GMT+0000 (UTC)

results matching ""

    No results matching ""