nv_dfm_core.session.ManualDispatcher

class nv_dfm_core.session.ManualDispatcher

Dispatcher that queues tokens for manual processing.

Tokens are queued and the user must periodically call: - job.callback_runner.process_pending() for sync callbacks - await job.callback_runner.process_pending_async() for async callbacks

This is useful for applications like Kit that need callbacks to run on the main thread.

Example

# Setup session = Session(

federation_name=”my_fed”, homesite=”site1”, callback_dispatcher=ManualDispatcher()

) job = session.execute(pipeline, default_callback=my_callback)

# In your main thread’s update loop: def on_update():

runner = job.callback_runner if runner and runner.has_pending():

runner.process_pending()

create_runner(default_callback=None, place_callbacks=None)

Create a ManualCallbackRunner for queued callback execution.

Parameters:

• default_callback (DfmDataCallback | None) – Fallback callback for tokens without a place-specific callback.

• place_callbacks (dict[str, DfmDataCallback] | None) – Dictionary mapping place names to callbacks.

Returns:

A ManualCallbackRunner instance with a token queue.

Return type:

ManualCallbackRunner