Allow asynchronous behaviour for long running tasks

Because zarpaint handles large datasets, some of the computations can be long running. Making these tasks asynchronous would improve useability, since the napari viewer would no longer freeze until the computation is complete.

A solution could be to use the napari @thread_worker decorator, ideally in combination with a progress bar so the user will know when the computation is complete.

The napari threading guide shows this example:

import napari
import numpy as np

from napari.qt.threading import thread_worker

@thread_worker
def average_large_image():
    return np.random.rand(1024, 512, 512).mean(0)

viewer = napari.Viewer()
worker = average_large_image()  # create "worker" object
worker.returned.connect(viewer.add_image)  # connect callback functions
worker.start()  # start the thread!
napari.run()

And Talley explains several different ways to update the progress bar from a thread worker in this thread on forum.sc:

@thread_worker(progress={"total": 10})
def my_func():
    ...

# or you can use:
def make_worker():
    n = figure_out_number_of_iterations()
    work = thread_worker(my_func, progress={"total": n})
    ...

# edit you can also use napari.qt.create_worker
from napari.qt import create_worker

def long_function(duration):
    import time
    time.sleep(duration)

worker = create_worker(long_function, 10, _progress={'total', n})
worker.start()

jni / zarpaint

Allow asynchronous behaviour for long running tasks #42